* [PATCH net-next 11/16] sfc: Refactor Falcon-arch filter removal
From: Ben Hutchings @ 2013-08-27 20:48 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377635844.13272.69.camel@bwh-desktop.uk.level5networks.com>
Move the special case for removal of default filters from
efx_farch_filter_table_clear_entry() into a wrapper function,
efx_farch_filter_table_remove(). Move the existence and priority
checks into the latter and use it where appropriate.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/farch.c | 74 ++++++++++++++++++++++------------------
1 file changed, 40 insertions(+), 34 deletions(-)
diff --git a/drivers/net/ethernet/sfc/farch.c b/drivers/net/ethernet/sfc/farch.c
index 1941804..d91bced 100644
--- a/drivers/net/ethernet/sfc/farch.c
+++ b/drivers/net/ethernet/sfc/farch.c
@@ -2516,33 +2516,49 @@ efx_farch_filter_table_clear_entry(struct efx_nic *efx,
{
static efx_oword_t filter;
+ EFX_WARN_ON_PARANOID(!test_bit(filter_idx, table->used_bitmap));
+
+ __clear_bit(filter_idx, table->used_bitmap);
+ --table->used;
+ memset(&table->spec[filter_idx], 0, sizeof(table->spec[0]));
+
+ efx_writeo(efx, &filter, table->offset + table->step * filter_idx);
+
+ /* If this filter required a greater search depth than
+ * any other, the search limit for its type can now be
+ * decreased. However, it is hard to determine that
+ * unless the table has become completely empty - in
+ * which case, all its search limits can be set to 0.
+ */
+ if (unlikely(table->used == 0)) {
+ memset(table->search_limit, 0, sizeof(table->search_limit));
+ if (table->id == EFX_FARCH_FILTER_TABLE_TX_MAC)
+ efx_farch_filter_push_tx_limits(efx);
+ else
+ efx_farch_filter_push_rx_config(efx);
+ }
+}
+
+static int efx_farch_filter_remove(struct efx_nic *efx,
+ struct efx_farch_filter_table *table,
+ unsigned int filter_idx,
+ enum efx_filter_priority priority)
+{
+ struct efx_farch_filter_spec *spec = &table->spec[filter_idx];
+
+ if (!test_bit(filter_idx, table->used_bitmap) ||
+ spec->priority > priority)
+ return -ENOENT;
+
if (table->id == EFX_FARCH_FILTER_TABLE_RX_DEF) {
/* RX default filters must always exist */
efx_farch_filter_reset_rx_def(efx, filter_idx);
efx_farch_filter_push_rx_config(efx);
- } else if (test_bit(filter_idx, table->used_bitmap)) {
- __clear_bit(filter_idx, table->used_bitmap);
- --table->used;
- memset(&table->spec[filter_idx], 0, sizeof(table->spec[0]));
-
- efx_writeo(efx, &filter,
- table->offset + table->step * filter_idx);
-
- /* If this filter required a greater search depth than
- * any other, the search limit for its type can now be
- * decreased. However, it is hard to determine that
- * unless the table has become completely empty - in
- * which case, all its search limits can be set to 0.
- */
- if (unlikely(table->used == 0)) {
- memset(table->search_limit, 0,
- sizeof(table->search_limit));
- if (table->id == EFX_FARCH_FILTER_TABLE_TX_MAC)
- efx_farch_filter_push_tx_limits(efx);
- else
- efx_farch_filter_push_rx_config(efx);
- }
+ } else {
+ efx_farch_filter_table_clear_entry(efx, table, filter_idx);
}
+
+ return 0;
}
int efx_farch_filter_remove_safe(struct efx_nic *efx,
@@ -2567,15 +2583,7 @@ int efx_farch_filter_remove_safe(struct efx_nic *efx,
spec = &table->spec[filter_idx];
spin_lock_bh(&efx->filter_lock);
-
- if (test_bit(filter_idx, table->used_bitmap) &&
- spec->priority == priority) {
- efx_farch_filter_table_clear_entry(efx, table, filter_idx);
- rc = 0;
- } else {
- rc = -ENOENT;
- }
-
+ rc = efx_farch_filter_remove(efx, table, filter_idx, priority);
spin_unlock_bh(&efx->filter_lock);
return rc;
@@ -2628,9 +2636,7 @@ efx_farch_filter_table_clear(struct efx_nic *efx,
spin_lock_bh(&efx->filter_lock);
for (filter_idx = 0; filter_idx < table->size; ++filter_idx)
- if (table->spec[filter_idx].priority <= priority)
- efx_farch_filter_table_clear_entry(efx, table,
- filter_idx);
+ efx_farch_filter_remove(efx, table, filter_idx, priority);
spin_unlock_bh(&efx->filter_lock);
}
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [PATCH net-next 10/16] sfc: Make most filter operations NIC-type-specific
From: Ben Hutchings @ 2013-08-27 20:48 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377635844.13272.69.camel@bwh-desktop.uk.level5networks.com>
Aside from accelerated RFS, there is almost nothing that can be shared
between the filter table implementations for the Falcon architecture
and EF10.
Move the few shared functions into efx.c and rx.c and the rest into
farch.c. Introduce efx_nic_type operations for the implementation and
inline wrapper functions that call these.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
This looks like a huge change, but is mostly moving functions without
changing their implementation.
Ben.
drivers/net/ethernet/sfc/Makefile | 1 -
drivers/net/ethernet/sfc/efx.c | 40 +-
drivers/net/ethernet/sfc/efx.h | 107 ++-
drivers/net/ethernet/sfc/falcon.c | 31 +
drivers/net/ethernet/sfc/farch.c | 1108 +++++++++++++++++++++++++++++
drivers/net/ethernet/sfc/filter.c | 1244 ---------------------------------
drivers/net/ethernet/sfc/net_driver.h | 46 ++
drivers/net/ethernet/sfc/nic.h | 28 +
drivers/net/ethernet/sfc/rx.c | 94 +++
drivers/net/ethernet/sfc/siena.c | 16 +
10 files changed, 1448 insertions(+), 1267 deletions(-)
delete mode 100644 drivers/net/ethernet/sfc/filter.c
diff --git a/drivers/net/ethernet/sfc/Makefile b/drivers/net/ethernet/sfc/Makefile
index ef7410f..a612726 100644
--- a/drivers/net/ethernet/sfc/Makefile
+++ b/drivers/net/ethernet/sfc/Makefile
@@ -1,5 +1,4 @@
sfc-y += efx.o nic.o farch.o falcon.o siena.o tx.o rx.o \
- filter.o \
selftest.o ethtool.o qt202x_phy.o mdio_10g.o \
tenxpress.o txc43128_phy.o falcon_boards.o \
mcdi.o mcdi_port.o mcdi_mon.o ptp.o
diff --git a/drivers/net/ethernet/sfc/efx.c b/drivers/net/ethernet/sfc/efx.c
index 49d06ca..a2daaae 100644
--- a/drivers/net/ethernet/sfc/efx.c
+++ b/drivers/net/ethernet/sfc/efx.c
@@ -611,7 +611,7 @@ static void efx_start_datapath(struct efx_nic *efx)
/* RX filters also have scatter-enabled flags */
if (efx->rx_scatter != old_rx_scatter)
- efx_filter_update_rx_scatter(efx);
+ efx->type->filter_update_rx_scatter(efx);
/* We must keep at least one descriptor in a TX ring empty.
* We could avoid this when the queue size does not exactly
@@ -1499,6 +1499,44 @@ static void efx_remove_nic(struct efx_nic *efx)
efx->type->remove(efx);
}
+static int efx_probe_filters(struct efx_nic *efx)
+{
+ int rc;
+
+ spin_lock_init(&efx->filter_lock);
+
+ rc = efx->type->filter_table_probe(efx);
+ if (rc)
+ return rc;
+
+#ifdef CONFIG_RFS_ACCEL
+ if (efx->type->offload_features & NETIF_F_NTUPLE) {
+ efx->rps_flow_id = kcalloc(efx->type->max_rx_ip_filters,
+ sizeof(*efx->rps_flow_id),
+ GFP_KERNEL);
+ if (!efx->rps_flow_id) {
+ efx->type->filter_table_remove(efx);
+ return -ENOMEM;
+ }
+ }
+#endif
+
+ return 0;
+}
+
+static void efx_remove_filters(struct efx_nic *efx)
+{
+#ifdef CONFIG_RFS_ACCEL
+ kfree(efx->rps_flow_id);
+#endif
+ efx->type->filter_table_remove(efx);
+}
+
+static void efx_restore_filters(struct efx_nic *efx)
+{
+ efx->type->filter_table_restore(efx);
+}
+
/**************************************************************************
*
* NIC startup/shutdown
diff --git a/drivers/net/ethernet/sfc/efx.h b/drivers/net/ethernet/sfc/efx.h
index 45de5b9..9e35738 100644
--- a/drivers/net/ethernet/sfc/efx.h
+++ b/drivers/net/ethernet/sfc/efx.h
@@ -68,27 +68,92 @@ extern void efx_schedule_slow_fill(struct efx_rx_queue *rx_queue);
#define EFX_TXQ_MIN_ENT(efx) (2 * efx_tx_max_skb_descs(efx))
/* Filters */
-extern int efx_probe_filters(struct efx_nic *efx);
-extern void efx_restore_filters(struct efx_nic *efx);
-extern void efx_remove_filters(struct efx_nic *efx);
-extern void efx_filter_update_rx_scatter(struct efx_nic *efx);
-extern s32 efx_filter_insert_filter(struct efx_nic *efx,
- struct efx_filter_spec *spec,
- bool replace);
-extern int efx_filter_remove_id_safe(struct efx_nic *efx,
- enum efx_filter_priority priority,
- u32 filter_id);
-extern int efx_filter_get_filter_safe(struct efx_nic *efx,
- enum efx_filter_priority priority,
- u32 filter_id, struct efx_filter_spec *);
-extern void efx_filter_clear_rx(struct efx_nic *efx,
- enum efx_filter_priority priority);
-extern u32 efx_filter_count_rx_used(struct efx_nic *efx,
- enum efx_filter_priority priority);
-extern u32 efx_filter_get_rx_id_limit(struct efx_nic *efx);
-extern s32 efx_filter_get_rx_ids(struct efx_nic *efx,
- enum efx_filter_priority priority,
- u32 *buf, u32 size);
+
+/**
+ * efx_filter_insert_filter - add or replace a filter
+ * @efx: NIC in which to insert the filter
+ * @spec: Specification for the filter
+ * @replace_equal: Flag for whether the specified filter may replace an
+ * existing filter with equal priority
+ *
+ * On success, return the filter ID.
+ * On failure, return a negative error code.
+ *
+ * If an existing filter has equal match values to the new filter
+ * spec, then the new filter might replace it, depending on the
+ * relative priorities. If the existing filter has lower priority, or
+ * if @replace_equal is set and it has equal priority, then it is
+ * replaced. Otherwise the function fails, returning -%EPERM if
+ * the existing filter has higher priority or -%EEXIST if it has
+ * equal priority.
+ */
+static inline s32 efx_filter_insert_filter(struct efx_nic *efx,
+ struct efx_filter_spec *spec,
+ bool replace_equal)
+{
+ return efx->type->filter_insert(efx, spec, replace_equal);
+}
+
+/**
+ * efx_filter_remove_id_safe - remove a filter by ID, carefully
+ * @efx: NIC from which to remove the filter
+ * @priority: Priority of filter, as passed to @efx_filter_insert_filter
+ * @filter_id: ID of filter, as returned by @efx_filter_insert_filter
+ *
+ * This function will range-check @filter_id, so it is safe to call
+ * with a value passed from userland.
+ */
+static inline int efx_filter_remove_id_safe(struct efx_nic *efx,
+ enum efx_filter_priority priority,
+ u32 filter_id)
+{
+ return efx->type->filter_remove_safe(efx, priority, filter_id);
+}
+
+/**
+ * efx_filter_get_filter_safe - retrieve a filter by ID, carefully
+ * @efx: NIC from which to remove the filter
+ * @priority: Priority of filter, as passed to @efx_filter_insert_filter
+ * @filter_id: ID of filter, as returned by @efx_filter_insert_filter
+ * @spec: Buffer in which to store filter specification
+ *
+ * This function will range-check @filter_id, so it is safe to call
+ * with a value passed from userland.
+ */
+static inline int
+efx_filter_get_filter_safe(struct efx_nic *efx,
+ enum efx_filter_priority priority,
+ u32 filter_id, struct efx_filter_spec *spec)
+{
+ return efx->type->filter_get_safe(efx, priority, filter_id, spec);
+}
+
+/**
+ * efx_farch_filter_clear_rx - remove RX filters by priority
+ * @efx: NIC from which to remove the filters
+ * @priority: Maximum priority to remove
+ */
+static inline void efx_filter_clear_rx(struct efx_nic *efx,
+ enum efx_filter_priority priority)
+{
+ return efx->type->filter_clear_rx(efx, priority);
+}
+
+static inline u32 efx_filter_count_rx_used(struct efx_nic *efx,
+ enum efx_filter_priority priority)
+{
+ return efx->type->filter_count_rx_used(efx, priority);
+}
+static inline u32 efx_filter_get_rx_id_limit(struct efx_nic *efx)
+{
+ return efx->type->filter_get_rx_id_limit(efx);
+}
+static inline s32 efx_filter_get_rx_ids(struct efx_nic *efx,
+ enum efx_filter_priority priority,
+ u32 *buf, u32 size)
+{
+ return efx->type->filter_get_rx_ids(efx, priority, buf, size);
+}
#ifdef CONFIG_RFS_ACCEL
extern int efx_filter_rfs(struct net_device *net_dev, const struct sk_buff *skb,
u16 rxq_index, u32 flow_id);
diff --git a/drivers/net/ethernet/sfc/falcon.c b/drivers/net/ethernet/sfc/falcon.c
index 0fd8a88..6ea28f8 100644
--- a/drivers/net/ethernet/sfc/falcon.c
+++ b/drivers/net/ethernet/sfc/falcon.c
@@ -2400,6 +2400,21 @@ const struct efx_nic_type falcon_a1_nic_type = {
.ev_read_ack = efx_farch_ev_read_ack,
.ev_test_generate = efx_farch_ev_test_generate,
+ /* We don't expose the filter table on Falcon A1 as it is not
+ * mapped into function 0, but these implementations still
+ * work with a degenerate case of all tables set to size 0.
+ */
+ .filter_table_probe = efx_farch_filter_table_probe,
+ .filter_table_restore = efx_farch_filter_table_restore,
+ .filter_table_remove = efx_farch_filter_table_remove,
+ .filter_insert = efx_farch_filter_insert,
+ .filter_remove_safe = efx_farch_filter_remove_safe,
+ .filter_get_safe = efx_farch_filter_get_safe,
+ .filter_clear_rx = efx_farch_filter_clear_rx,
+ .filter_count_rx_used = efx_farch_filter_count_rx_used,
+ .filter_get_rx_id_limit = efx_farch_filter_get_rx_id_limit,
+ .filter_get_rx_ids = efx_farch_filter_get_rx_ids,
+
.revision = EFX_REV_FALCON_A1,
.txd_ptr_tbl_base = FR_AA_TX_DESC_PTR_TBL_KER,
.rxd_ptr_tbl_base = FR_AA_RX_DESC_PTR_TBL_KER,
@@ -2468,6 +2483,21 @@ const struct efx_nic_type falcon_b0_nic_type = {
.ev_process = efx_farch_ev_process,
.ev_read_ack = efx_farch_ev_read_ack,
.ev_test_generate = efx_farch_ev_test_generate,
+ .filter_table_probe = efx_farch_filter_table_probe,
+ .filter_table_restore = efx_farch_filter_table_restore,
+ .filter_table_remove = efx_farch_filter_table_remove,
+ .filter_update_rx_scatter = efx_farch_filter_update_rx_scatter,
+ .filter_insert = efx_farch_filter_insert,
+ .filter_remove_safe = efx_farch_filter_remove_safe,
+ .filter_get_safe = efx_farch_filter_get_safe,
+ .filter_clear_rx = efx_farch_filter_clear_rx,
+ .filter_count_rx_used = efx_farch_filter_count_rx_used,
+ .filter_get_rx_id_limit = efx_farch_filter_get_rx_id_limit,
+ .filter_get_rx_ids = efx_farch_filter_get_rx_ids,
+#ifdef CONFIG_RFS_ACCEL
+ .filter_rfs_insert = efx_farch_filter_rfs_insert,
+ .filter_rfs_expire_one = efx_farch_filter_rfs_expire_one,
+#endif
.revision = EFX_REV_FALCON_B0,
.txd_ptr_tbl_base = FR_BZ_TX_DESC_PTR_TBL,
@@ -2483,5 +2513,6 @@ const struct efx_nic_type falcon_b0_nic_type = {
.timer_period_max = 1 << FRF_AB_TC_TIMER_VAL_WIDTH,
.offload_features = NETIF_F_IP_CSUM | NETIF_F_RXHASH | NETIF_F_NTUPLE,
.mcdi_max_ver = -1,
+ .max_rx_ip_filters = FR_BZ_RX_FILTER_TBL0_ROWS,
};
diff --git a/drivers/net/ethernet/sfc/farch.c b/drivers/net/ethernet/sfc/farch.c
index 7f50882..1941804 100644
--- a/drivers/net/ethernet/sfc/farch.c
+++ b/drivers/net/ethernet/sfc/farch.c
@@ -1779,3 +1779,1111 @@ void efx_farch_init_common(struct efx_nic *efx)
efx_writeo(efx, &temp, FR_BZ_TX_PACE);
}
}
+
+/**************************************************************************
+ *
+ * Filter tables
+ *
+ **************************************************************************
+ */
+
+/* "Fudge factors" - difference between programmed value and actual depth.
+ * Due to pipelined implementation we need to program H/W with a value that
+ * is larger than the hop limit we want.
+ */
+#define EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD 3
+#define EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL 1
+
+/* Hard maximum search limit. Hardware will time-out beyond 200-something.
+ * We also need to avoid infinite loops in efx_farch_filter_search() when the
+ * table is full.
+ */
+#define EFX_FARCH_FILTER_CTL_SRCH_MAX 200
+
+/* Don't try very hard to find space for performance hints, as this is
+ * counter-productive. */
+#define EFX_FARCH_FILTER_CTL_SRCH_HINT_MAX 5
+
+enum efx_farch_filter_type {
+ EFX_FARCH_FILTER_TCP_FULL = 0,
+ EFX_FARCH_FILTER_TCP_WILD,
+ EFX_FARCH_FILTER_UDP_FULL,
+ EFX_FARCH_FILTER_UDP_WILD,
+ EFX_FARCH_FILTER_MAC_FULL = 4,
+ EFX_FARCH_FILTER_MAC_WILD,
+ EFX_FARCH_FILTER_UC_DEF = 8,
+ EFX_FARCH_FILTER_MC_DEF,
+ EFX_FARCH_FILTER_TYPE_COUNT, /* number of specific types */
+};
+
+enum efx_farch_filter_table_id {
+ EFX_FARCH_FILTER_TABLE_RX_IP = 0,
+ EFX_FARCH_FILTER_TABLE_RX_MAC,
+ EFX_FARCH_FILTER_TABLE_RX_DEF,
+ EFX_FARCH_FILTER_TABLE_TX_MAC,
+ EFX_FARCH_FILTER_TABLE_COUNT,
+};
+
+enum efx_farch_filter_index {
+ EFX_FARCH_FILTER_INDEX_UC_DEF,
+ EFX_FARCH_FILTER_INDEX_MC_DEF,
+ EFX_FARCH_FILTER_SIZE_RX_DEF,
+};
+
+struct efx_farch_filter_spec {
+ u8 type:4;
+ u8 priority:4;
+ u8 flags;
+ u16 dmaq_id;
+ u32 data[3];
+};
+
+struct efx_farch_filter_table {
+ enum efx_farch_filter_table_id id;
+ u32 offset; /* address of table relative to BAR */
+ unsigned size; /* number of entries */
+ unsigned step; /* step between entries */
+ unsigned used; /* number currently used */
+ unsigned long *used_bitmap;
+ struct efx_farch_filter_spec *spec;
+ unsigned search_limit[EFX_FARCH_FILTER_TYPE_COUNT];
+};
+
+struct efx_farch_filter_state {
+ struct efx_farch_filter_table table[EFX_FARCH_FILTER_TABLE_COUNT];
+};
+
+static void
+efx_farch_filter_table_clear_entry(struct efx_nic *efx,
+ struct efx_farch_filter_table *table,
+ unsigned int filter_idx);
+
+/* The filter hash function is LFSR polynomial x^16 + x^3 + 1 of a 32-bit
+ * key derived from the n-tuple. The initial LFSR state is 0xffff. */
+static u16 efx_farch_filter_hash(u32 key)
+{
+ u16 tmp;
+
+ /* First 16 rounds */
+ tmp = 0x1fff ^ key >> 16;
+ tmp = tmp ^ tmp >> 3 ^ tmp >> 6;
+ tmp = tmp ^ tmp >> 9;
+ /* Last 16 rounds */
+ tmp = tmp ^ tmp << 13 ^ key;
+ tmp = tmp ^ tmp >> 3 ^ tmp >> 6;
+ return tmp ^ tmp >> 9;
+}
+
+/* To allow for hash collisions, filter search continues at these
+ * increments from the first possible entry selected by the hash. */
+static u16 efx_farch_filter_increment(u32 key)
+{
+ return key * 2 - 1;
+}
+
+static enum efx_farch_filter_table_id
+efx_farch_filter_spec_table_id(const struct efx_farch_filter_spec *spec)
+{
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
+ (EFX_FARCH_FILTER_TCP_FULL >> 2));
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
+ (EFX_FARCH_FILTER_TCP_WILD >> 2));
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
+ (EFX_FARCH_FILTER_UDP_FULL >> 2));
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
+ (EFX_FARCH_FILTER_UDP_WILD >> 2));
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_MAC !=
+ (EFX_FARCH_FILTER_MAC_FULL >> 2));
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_MAC !=
+ (EFX_FARCH_FILTER_MAC_WILD >> 2));
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_TX_MAC !=
+ EFX_FARCH_FILTER_TABLE_RX_MAC + 2);
+ return (spec->type >> 2) + ((spec->flags & EFX_FILTER_FLAG_TX) ? 2 : 0);
+}
+
+static void efx_farch_filter_push_rx_config(struct efx_nic *efx)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_table *table;
+ efx_oword_t filter_ctl;
+
+ efx_reado(efx, &filter_ctl, FR_BZ_RX_FILTER_CTL);
+
+ table = &state->table[EFX_FARCH_FILTER_TABLE_RX_IP];
+ EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_TCP_FULL_SRCH_LIMIT,
+ table->search_limit[EFX_FARCH_FILTER_TCP_FULL] +
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
+ EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_TCP_WILD_SRCH_LIMIT,
+ table->search_limit[EFX_FARCH_FILTER_TCP_WILD] +
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
+ EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_UDP_FULL_SRCH_LIMIT,
+ table->search_limit[EFX_FARCH_FILTER_UDP_FULL] +
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
+ EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_UDP_WILD_SRCH_LIMIT,
+ table->search_limit[EFX_FARCH_FILTER_UDP_WILD] +
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
+
+ table = &state->table[EFX_FARCH_FILTER_TABLE_RX_MAC];
+ if (table->size) {
+ EFX_SET_OWORD_FIELD(
+ filter_ctl, FRF_CZ_ETHERNET_FULL_SEARCH_LIMIT,
+ table->search_limit[EFX_FARCH_FILTER_MAC_FULL] +
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
+ EFX_SET_OWORD_FIELD(
+ filter_ctl, FRF_CZ_ETHERNET_WILDCARD_SEARCH_LIMIT,
+ table->search_limit[EFX_FARCH_FILTER_MAC_WILD] +
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
+ }
+
+ table = &state->table[EFX_FARCH_FILTER_TABLE_RX_DEF];
+ if (table->size) {
+ EFX_SET_OWORD_FIELD(
+ filter_ctl, FRF_CZ_UNICAST_NOMATCH_Q_ID,
+ table->spec[EFX_FARCH_FILTER_INDEX_UC_DEF].dmaq_id);
+ EFX_SET_OWORD_FIELD(
+ filter_ctl, FRF_CZ_UNICAST_NOMATCH_RSS_ENABLED,
+ !!(table->spec[EFX_FARCH_FILTER_INDEX_UC_DEF].flags &
+ EFX_FILTER_FLAG_RX_RSS));
+ EFX_SET_OWORD_FIELD(
+ filter_ctl, FRF_CZ_MULTICAST_NOMATCH_Q_ID,
+ table->spec[EFX_FARCH_FILTER_INDEX_MC_DEF].dmaq_id);
+ EFX_SET_OWORD_FIELD(
+ filter_ctl, FRF_CZ_MULTICAST_NOMATCH_RSS_ENABLED,
+ !!(table->spec[EFX_FARCH_FILTER_INDEX_MC_DEF].flags &
+ EFX_FILTER_FLAG_RX_RSS));
+
+ /* There is a single bit to enable RX scatter for all
+ * unmatched packets. Only set it if scatter is
+ * enabled in both filter specs.
+ */
+ EFX_SET_OWORD_FIELD(
+ filter_ctl, FRF_BZ_SCATTER_ENBL_NO_MATCH_Q,
+ !!(table->spec[EFX_FARCH_FILTER_INDEX_UC_DEF].flags &
+ table->spec[EFX_FARCH_FILTER_INDEX_MC_DEF].flags &
+ EFX_FILTER_FLAG_RX_SCATTER));
+ } else if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0) {
+ /* We don't expose 'default' filters because unmatched
+ * packets always go to the queue number found in the
+ * RSS table. But we still need to set the RX scatter
+ * bit here.
+ */
+ EFX_SET_OWORD_FIELD(
+ filter_ctl, FRF_BZ_SCATTER_ENBL_NO_MATCH_Q,
+ efx->rx_scatter);
+ }
+
+ efx_writeo(efx, &filter_ctl, FR_BZ_RX_FILTER_CTL);
+}
+
+static void efx_farch_filter_push_tx_limits(struct efx_nic *efx)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_table *table;
+ efx_oword_t tx_cfg;
+
+ efx_reado(efx, &tx_cfg, FR_AZ_TX_CFG);
+
+ table = &state->table[EFX_FARCH_FILTER_TABLE_TX_MAC];
+ if (table->size) {
+ EFX_SET_OWORD_FIELD(
+ tx_cfg, FRF_CZ_TX_ETH_FILTER_FULL_SEARCH_RANGE,
+ table->search_limit[EFX_FARCH_FILTER_MAC_FULL] +
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
+ EFX_SET_OWORD_FIELD(
+ tx_cfg, FRF_CZ_TX_ETH_FILTER_WILD_SEARCH_RANGE,
+ table->search_limit[EFX_FARCH_FILTER_MAC_WILD] +
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
+ }
+
+ efx_writeo(efx, &tx_cfg, FR_AZ_TX_CFG);
+}
+
+static int
+efx_farch_filter_from_gen_spec(struct efx_farch_filter_spec *spec,
+ const struct efx_filter_spec *gen_spec)
+{
+ bool is_full = false;
+
+ if ((gen_spec->flags & EFX_FILTER_FLAG_RX_RSS) &&
+ gen_spec->rss_context != EFX_FILTER_RSS_CONTEXT_DEFAULT)
+ return -EINVAL;
+
+ spec->priority = gen_spec->priority;
+ spec->flags = gen_spec->flags;
+ spec->dmaq_id = gen_spec->dmaq_id;
+
+ switch (gen_spec->match_flags) {
+ case (EFX_FILTER_MATCH_ETHER_TYPE | EFX_FILTER_MATCH_IP_PROTO |
+ EFX_FILTER_MATCH_LOC_HOST | EFX_FILTER_MATCH_LOC_PORT |
+ EFX_FILTER_MATCH_REM_HOST | EFX_FILTER_MATCH_REM_PORT):
+ is_full = true;
+ /* fall through */
+ case (EFX_FILTER_MATCH_ETHER_TYPE | EFX_FILTER_MATCH_IP_PROTO |
+ EFX_FILTER_MATCH_LOC_HOST | EFX_FILTER_MATCH_LOC_PORT): {
+ __be32 rhost, host1, host2;
+ __be16 rport, port1, port2;
+
+ EFX_BUG_ON_PARANOID(!(gen_spec->flags & EFX_FILTER_FLAG_RX));
+
+ if (gen_spec->ether_type != htons(ETH_P_IP))
+ return -EPROTONOSUPPORT;
+ if (gen_spec->loc_port == 0 ||
+ (is_full && gen_spec->rem_port == 0))
+ return -EADDRNOTAVAIL;
+ switch (gen_spec->ip_proto) {
+ case IPPROTO_TCP:
+ spec->type = (is_full ? EFX_FARCH_FILTER_TCP_FULL :
+ EFX_FARCH_FILTER_TCP_WILD);
+ break;
+ case IPPROTO_UDP:
+ spec->type = (is_full ? EFX_FARCH_FILTER_UDP_FULL :
+ EFX_FARCH_FILTER_UDP_WILD);
+ break;
+ default:
+ return -EPROTONOSUPPORT;
+ }
+
+ /* Filter is constructed in terms of source and destination,
+ * with the odd wrinkle that the ports are swapped in a UDP
+ * wildcard filter. We need to convert from local and remote
+ * (= zero for wildcard) addresses.
+ */
+ rhost = is_full ? gen_spec->rem_host[0] : 0;
+ rport = is_full ? gen_spec->rem_port : 0;
+ host1 = rhost;
+ host2 = gen_spec->loc_host[0];
+ if (!is_full && gen_spec->ip_proto == IPPROTO_UDP) {
+ port1 = gen_spec->loc_port;
+ port2 = rport;
+ } else {
+ port1 = rport;
+ port2 = gen_spec->loc_port;
+ }
+ spec->data[0] = ntohl(host1) << 16 | ntohs(port1);
+ spec->data[1] = ntohs(port2) << 16 | ntohl(host1) >> 16;
+ spec->data[2] = ntohl(host2);
+
+ break;
+ }
+
+ case EFX_FILTER_MATCH_LOC_MAC | EFX_FILTER_MATCH_OUTER_VID:
+ is_full = true;
+ /* fall through */
+ case EFX_FILTER_MATCH_LOC_MAC:
+ spec->type = (is_full ? EFX_FARCH_FILTER_MAC_FULL :
+ EFX_FARCH_FILTER_MAC_WILD);
+ spec->data[0] = is_full ? ntohs(gen_spec->outer_vid) : 0;
+ spec->data[1] = (gen_spec->loc_mac[2] << 24 |
+ gen_spec->loc_mac[3] << 16 |
+ gen_spec->loc_mac[4] << 8 |
+ gen_spec->loc_mac[5]);
+ spec->data[2] = (gen_spec->loc_mac[0] << 8 |
+ gen_spec->loc_mac[1]);
+ break;
+
+ case EFX_FILTER_MATCH_LOC_MAC_IG:
+ spec->type = (is_multicast_ether_addr(gen_spec->loc_mac) ?
+ EFX_FARCH_FILTER_MC_DEF :
+ EFX_FARCH_FILTER_UC_DEF);
+ memset(spec->data, 0, sizeof(spec->data)); /* ensure equality */
+ break;
+
+ default:
+ return -EPROTONOSUPPORT;
+ }
+
+ return 0;
+}
+
+static void
+efx_farch_filter_to_gen_spec(struct efx_filter_spec *gen_spec,
+ const struct efx_farch_filter_spec *spec)
+{
+ bool is_full = false;
+
+ /* *gen_spec should be completely initialised, to be consistent
+ * with efx_filter_init_{rx,tx}() and in case we want to copy
+ * it back to userland.
+ */
+ memset(gen_spec, 0, sizeof(*gen_spec));
+
+ gen_spec->priority = spec->priority;
+ gen_spec->flags = spec->flags;
+ gen_spec->dmaq_id = spec->dmaq_id;
+
+ switch (spec->type) {
+ case EFX_FARCH_FILTER_TCP_FULL:
+ case EFX_FARCH_FILTER_UDP_FULL:
+ is_full = true;
+ /* fall through */
+ case EFX_FARCH_FILTER_TCP_WILD:
+ case EFX_FARCH_FILTER_UDP_WILD: {
+ __be32 host1, host2;
+ __be16 port1, port2;
+
+ gen_spec->match_flags =
+ EFX_FILTER_MATCH_ETHER_TYPE |
+ EFX_FILTER_MATCH_IP_PROTO |
+ EFX_FILTER_MATCH_LOC_HOST | EFX_FILTER_MATCH_LOC_PORT;
+ if (is_full)
+ gen_spec->match_flags |= (EFX_FILTER_MATCH_REM_HOST |
+ EFX_FILTER_MATCH_REM_PORT);
+ gen_spec->ether_type = htons(ETH_P_IP);
+ gen_spec->ip_proto =
+ (spec->type == EFX_FARCH_FILTER_TCP_FULL ||
+ spec->type == EFX_FARCH_FILTER_TCP_WILD) ?
+ IPPROTO_TCP : IPPROTO_UDP;
+
+ host1 = htonl(spec->data[0] >> 16 | spec->data[1] << 16);
+ port1 = htons(spec->data[0]);
+ host2 = htonl(spec->data[2]);
+ port2 = htons(spec->data[1] >> 16);
+ if (spec->flags & EFX_FILTER_FLAG_TX) {
+ gen_spec->loc_host[0] = host1;
+ gen_spec->rem_host[0] = host2;
+ } else {
+ gen_spec->loc_host[0] = host2;
+ gen_spec->rem_host[0] = host1;
+ }
+ if (!!(gen_spec->flags & EFX_FILTER_FLAG_TX) ^
+ (!is_full && gen_spec->ip_proto == IPPROTO_UDP)) {
+ gen_spec->loc_port = port1;
+ gen_spec->rem_port = port2;
+ } else {
+ gen_spec->loc_port = port2;
+ gen_spec->rem_port = port1;
+ }
+
+ break;
+ }
+
+ case EFX_FARCH_FILTER_MAC_FULL:
+ is_full = true;
+ /* fall through */
+ case EFX_FARCH_FILTER_MAC_WILD:
+ gen_spec->match_flags = EFX_FILTER_MATCH_LOC_MAC;
+ if (is_full)
+ gen_spec->match_flags |= EFX_FILTER_MATCH_OUTER_VID;
+ gen_spec->loc_mac[0] = spec->data[2] >> 8;
+ gen_spec->loc_mac[1] = spec->data[2];
+ gen_spec->loc_mac[2] = spec->data[1] >> 24;
+ gen_spec->loc_mac[3] = spec->data[1] >> 16;
+ gen_spec->loc_mac[4] = spec->data[1] >> 8;
+ gen_spec->loc_mac[5] = spec->data[1];
+ gen_spec->outer_vid = htons(spec->data[0]);
+ break;
+
+ case EFX_FARCH_FILTER_UC_DEF:
+ case EFX_FARCH_FILTER_MC_DEF:
+ gen_spec->match_flags = EFX_FILTER_MATCH_LOC_MAC_IG;
+ gen_spec->loc_mac[0] = spec->type == EFX_FARCH_FILTER_MC_DEF;
+ break;
+
+ default:
+ WARN_ON(1);
+ break;
+ }
+}
+
+static void
+efx_farch_filter_reset_rx_def(struct efx_nic *efx, unsigned filter_idx)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_table *table =
+ &state->table[EFX_FARCH_FILTER_TABLE_RX_DEF];
+ struct efx_farch_filter_spec *spec = &table->spec[filter_idx];
+
+ /* If there's only one channel then disable RSS for non VF
+ * traffic, thereby allowing VFs to use RSS when the PF can't.
+ */
+ spec->type = EFX_FARCH_FILTER_UC_DEF + filter_idx;
+ spec->priority = EFX_FILTER_PRI_MANUAL;
+ spec->flags = (EFX_FILTER_FLAG_RX |
+ (efx->n_rx_channels > 1 ? EFX_FILTER_FLAG_RX_RSS : 0) |
+ (efx->rx_scatter ? EFX_FILTER_FLAG_RX_SCATTER : 0));
+ spec->dmaq_id = 0;
+ table->used_bitmap[0] |= 1 << filter_idx;
+}
+
+/* Build a filter entry and return its n-tuple key. */
+static u32 efx_farch_filter_build(efx_oword_t *filter,
+ struct efx_farch_filter_spec *spec)
+{
+ u32 data3;
+
+ switch (efx_farch_filter_spec_table_id(spec)) {
+ case EFX_FARCH_FILTER_TABLE_RX_IP: {
+ bool is_udp = (spec->type == EFX_FARCH_FILTER_UDP_FULL ||
+ spec->type == EFX_FARCH_FILTER_UDP_WILD);
+ EFX_POPULATE_OWORD_7(
+ *filter,
+ FRF_BZ_RSS_EN,
+ !!(spec->flags & EFX_FILTER_FLAG_RX_RSS),
+ FRF_BZ_SCATTER_EN,
+ !!(spec->flags & EFX_FILTER_FLAG_RX_SCATTER),
+ FRF_BZ_TCP_UDP, is_udp,
+ FRF_BZ_RXQ_ID, spec->dmaq_id,
+ EFX_DWORD_2, spec->data[2],
+ EFX_DWORD_1, spec->data[1],
+ EFX_DWORD_0, spec->data[0]);
+ data3 = is_udp;
+ break;
+ }
+
+ case EFX_FARCH_FILTER_TABLE_RX_MAC: {
+ bool is_wild = spec->type == EFX_FARCH_FILTER_MAC_WILD;
+ EFX_POPULATE_OWORD_7(
+ *filter,
+ FRF_CZ_RMFT_RSS_EN,
+ !!(spec->flags & EFX_FILTER_FLAG_RX_RSS),
+ FRF_CZ_RMFT_SCATTER_EN,
+ !!(spec->flags & EFX_FILTER_FLAG_RX_SCATTER),
+ FRF_CZ_RMFT_RXQ_ID, spec->dmaq_id,
+ FRF_CZ_RMFT_WILDCARD_MATCH, is_wild,
+ FRF_CZ_RMFT_DEST_MAC_HI, spec->data[2],
+ FRF_CZ_RMFT_DEST_MAC_LO, spec->data[1],
+ FRF_CZ_RMFT_VLAN_ID, spec->data[0]);
+ data3 = is_wild;
+ break;
+ }
+
+ case EFX_FARCH_FILTER_TABLE_TX_MAC: {
+ bool is_wild = spec->type == EFX_FARCH_FILTER_MAC_WILD;
+ EFX_POPULATE_OWORD_5(*filter,
+ FRF_CZ_TMFT_TXQ_ID, spec->dmaq_id,
+ FRF_CZ_TMFT_WILDCARD_MATCH, is_wild,
+ FRF_CZ_TMFT_SRC_MAC_HI, spec->data[2],
+ FRF_CZ_TMFT_SRC_MAC_LO, spec->data[1],
+ FRF_CZ_TMFT_VLAN_ID, spec->data[0]);
+ data3 = is_wild | spec->dmaq_id << 1;
+ break;
+ }
+
+ default:
+ BUG();
+ }
+
+ return spec->data[0] ^ spec->data[1] ^ spec->data[2] ^ data3;
+}
+
+static bool efx_farch_filter_equal(const struct efx_farch_filter_spec *left,
+ const struct efx_farch_filter_spec *right)
+{
+ if (left->type != right->type ||
+ memcmp(left->data, right->data, sizeof(left->data)))
+ return false;
+
+ if (left->flags & EFX_FILTER_FLAG_TX &&
+ left->dmaq_id != right->dmaq_id)
+ return false;
+
+ return true;
+}
+
+/*
+ * Construct/deconstruct external filter IDs. At least the RX filter
+ * IDs must be ordered by matching priority, for RX NFC semantics.
+ *
+ * Deconstruction needs to be robust against invalid IDs so that
+ * efx_filter_remove_id_safe() and efx_filter_get_filter_safe() can
+ * accept user-provided IDs.
+ */
+
+#define EFX_FARCH_FILTER_MATCH_PRI_COUNT 5
+
+static const u8 efx_farch_filter_type_match_pri[EFX_FARCH_FILTER_TYPE_COUNT] = {
+ [EFX_FARCH_FILTER_TCP_FULL] = 0,
+ [EFX_FARCH_FILTER_UDP_FULL] = 0,
+ [EFX_FARCH_FILTER_TCP_WILD] = 1,
+ [EFX_FARCH_FILTER_UDP_WILD] = 1,
+ [EFX_FARCH_FILTER_MAC_FULL] = 2,
+ [EFX_FARCH_FILTER_MAC_WILD] = 3,
+ [EFX_FARCH_FILTER_UC_DEF] = 4,
+ [EFX_FARCH_FILTER_MC_DEF] = 4,
+};
+
+static const enum efx_farch_filter_table_id efx_farch_filter_range_table[] = {
+ EFX_FARCH_FILTER_TABLE_RX_IP, /* RX match pri 0 */
+ EFX_FARCH_FILTER_TABLE_RX_IP,
+ EFX_FARCH_FILTER_TABLE_RX_MAC,
+ EFX_FARCH_FILTER_TABLE_RX_MAC,
+ EFX_FARCH_FILTER_TABLE_RX_DEF, /* RX match pri 4 */
+ EFX_FARCH_FILTER_TABLE_TX_MAC, /* TX match pri 0 */
+ EFX_FARCH_FILTER_TABLE_TX_MAC, /* TX match pri 1 */
+};
+
+#define EFX_FARCH_FILTER_INDEX_WIDTH 13
+#define EFX_FARCH_FILTER_INDEX_MASK ((1 << EFX_FARCH_FILTER_INDEX_WIDTH) - 1)
+
+static inline u32
+efx_farch_filter_make_id(const struct efx_farch_filter_spec *spec,
+ unsigned int index)
+{
+ unsigned int range;
+
+ range = efx_farch_filter_type_match_pri[spec->type];
+ if (!(spec->flags & EFX_FILTER_FLAG_RX))
+ range += EFX_FARCH_FILTER_MATCH_PRI_COUNT;
+
+ return range << EFX_FARCH_FILTER_INDEX_WIDTH | index;
+}
+
+static inline enum efx_farch_filter_table_id
+efx_farch_filter_id_table_id(u32 id)
+{
+ unsigned int range = id >> EFX_FARCH_FILTER_INDEX_WIDTH;
+
+ if (range < ARRAY_SIZE(efx_farch_filter_range_table))
+ return efx_farch_filter_range_table[range];
+ else
+ return EFX_FARCH_FILTER_TABLE_COUNT; /* invalid */
+}
+
+static inline unsigned int efx_farch_filter_id_index(u32 id)
+{
+ return id & EFX_FARCH_FILTER_INDEX_MASK;
+}
+
+u32 efx_farch_filter_get_rx_id_limit(struct efx_nic *efx)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ unsigned int range = EFX_FARCH_FILTER_MATCH_PRI_COUNT - 1;
+ enum efx_farch_filter_table_id table_id;
+
+ do {
+ table_id = efx_farch_filter_range_table[range];
+ if (state->table[table_id].size != 0)
+ return range << EFX_FARCH_FILTER_INDEX_WIDTH |
+ state->table[table_id].size;
+ } while (range--);
+
+ return 0;
+}
+
+s32 efx_farch_filter_insert(struct efx_nic *efx,
+ struct efx_filter_spec *gen_spec,
+ bool replace_equal)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_table *table;
+ struct efx_farch_filter_spec spec;
+ efx_oword_t filter;
+ int rep_index, ins_index;
+ unsigned int depth = 0;
+ int rc;
+
+ rc = efx_farch_filter_from_gen_spec(&spec, gen_spec);
+ if (rc)
+ return rc;
+
+ table = &state->table[efx_farch_filter_spec_table_id(&spec)];
+ if (table->size == 0)
+ return -EINVAL;
+
+ netif_vdbg(efx, hw, efx->net_dev,
+ "%s: type %d search_limit=%d", __func__, spec.type,
+ table->search_limit[spec.type]);
+
+ if (table->id == EFX_FARCH_FILTER_TABLE_RX_DEF) {
+ /* One filter spec per type */
+ BUILD_BUG_ON(EFX_FARCH_FILTER_INDEX_UC_DEF != 0);
+ BUILD_BUG_ON(EFX_FARCH_FILTER_INDEX_MC_DEF !=
+ EFX_FARCH_FILTER_MC_DEF - EFX_FARCH_FILTER_UC_DEF);
+ rep_index = spec.type - EFX_FARCH_FILTER_UC_DEF;
+ ins_index = rep_index;
+
+ spin_lock_bh(&efx->filter_lock);
+ } else {
+ /* Search concurrently for
+ * (1) a filter to be replaced (rep_index): any filter
+ * with the same match values, up to the current
+ * search depth for this type, and
+ * (2) the insertion point (ins_index): (1) or any
+ * free slot before it or up to the maximum search
+ * depth for this priority
+ * We fail if we cannot find (2).
+ *
+ * We can stop once either
+ * (a) we find (1), in which case we have definitely
+ * found (2) as well; or
+ * (b) we have searched exhaustively for (1), and have
+ * either found (2) or searched exhaustively for it
+ */
+ u32 key = efx_farch_filter_build(&filter, &spec);
+ unsigned int hash = efx_farch_filter_hash(key);
+ unsigned int incr = efx_farch_filter_increment(key);
+ unsigned int max_rep_depth = table->search_limit[spec.type];
+ unsigned int max_ins_depth =
+ spec.priority <= EFX_FILTER_PRI_HINT ?
+ EFX_FARCH_FILTER_CTL_SRCH_HINT_MAX :
+ EFX_FARCH_FILTER_CTL_SRCH_MAX;
+ unsigned int i = hash & (table->size - 1);
+
+ ins_index = -1;
+ depth = 1;
+
+ spin_lock_bh(&efx->filter_lock);
+
+ for (;;) {
+ if (!test_bit(i, table->used_bitmap)) {
+ if (ins_index < 0)
+ ins_index = i;
+ } else if (efx_farch_filter_equal(&spec,
+ &table->spec[i])) {
+ /* Case (a) */
+ if (ins_index < 0)
+ ins_index = i;
+ rep_index = i;
+ break;
+ }
+
+ if (depth >= max_rep_depth &&
+ (ins_index >= 0 || depth >= max_ins_depth)) {
+ /* Case (b) */
+ if (ins_index < 0) {
+ rc = -EBUSY;
+ goto out;
+ }
+ rep_index = -1;
+ break;
+ }
+
+ i = (i + incr) & (table->size - 1);
+ ++depth;
+ }
+ }
+
+ /* If we found a filter to be replaced, check whether we
+ * should do so
+ */
+ if (rep_index >= 0) {
+ struct efx_farch_filter_spec *saved_spec =
+ &table->spec[rep_index];
+
+ if (spec.priority == saved_spec->priority && !replace_equal) {
+ rc = -EEXIST;
+ goto out;
+ }
+ if (spec.priority < saved_spec->priority) {
+ rc = -EPERM;
+ goto out;
+ }
+ }
+
+ /* Insert the filter */
+ if (ins_index != rep_index) {
+ __set_bit(ins_index, table->used_bitmap);
+ ++table->used;
+ }
+ table->spec[ins_index] = spec;
+
+ if (table->id == EFX_FARCH_FILTER_TABLE_RX_DEF) {
+ efx_farch_filter_push_rx_config(efx);
+ } else {
+ if (table->search_limit[spec.type] < depth) {
+ table->search_limit[spec.type] = depth;
+ if (spec.flags & EFX_FILTER_FLAG_TX)
+ efx_farch_filter_push_tx_limits(efx);
+ else
+ efx_farch_filter_push_rx_config(efx);
+ }
+
+ efx_writeo(efx, &filter,
+ table->offset + table->step * ins_index);
+
+ /* If we were able to replace a filter by inserting
+ * at a lower depth, clear the replaced filter
+ */
+ if (ins_index != rep_index && rep_index >= 0)
+ efx_farch_filter_table_clear_entry(efx, table,
+ rep_index);
+ }
+
+ netif_vdbg(efx, hw, efx->net_dev,
+ "%s: filter type %d index %d rxq %u set",
+ __func__, spec.type, ins_index, spec.dmaq_id);
+ rc = efx_farch_filter_make_id(&spec, ins_index);
+
+out:
+ spin_unlock_bh(&efx->filter_lock);
+ return rc;
+}
+
+static void
+efx_farch_filter_table_clear_entry(struct efx_nic *efx,
+ struct efx_farch_filter_table *table,
+ unsigned int filter_idx)
+{
+ static efx_oword_t filter;
+
+ if (table->id == EFX_FARCH_FILTER_TABLE_RX_DEF) {
+ /* RX default filters must always exist */
+ efx_farch_filter_reset_rx_def(efx, filter_idx);
+ efx_farch_filter_push_rx_config(efx);
+ } else if (test_bit(filter_idx, table->used_bitmap)) {
+ __clear_bit(filter_idx, table->used_bitmap);
+ --table->used;
+ memset(&table->spec[filter_idx], 0, sizeof(table->spec[0]));
+
+ efx_writeo(efx, &filter,
+ table->offset + table->step * filter_idx);
+
+ /* If this filter required a greater search depth than
+ * any other, the search limit for its type can now be
+ * decreased. However, it is hard to determine that
+ * unless the table has become completely empty - in
+ * which case, all its search limits can be set to 0.
+ */
+ if (unlikely(table->used == 0)) {
+ memset(table->search_limit, 0,
+ sizeof(table->search_limit));
+ if (table->id == EFX_FARCH_FILTER_TABLE_TX_MAC)
+ efx_farch_filter_push_tx_limits(efx);
+ else
+ efx_farch_filter_push_rx_config(efx);
+ }
+ }
+}
+
+int efx_farch_filter_remove_safe(struct efx_nic *efx,
+ enum efx_filter_priority priority,
+ u32 filter_id)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ enum efx_farch_filter_table_id table_id;
+ struct efx_farch_filter_table *table;
+ unsigned int filter_idx;
+ struct efx_farch_filter_spec *spec;
+ int rc;
+
+ table_id = efx_farch_filter_id_table_id(filter_id);
+ if ((unsigned int)table_id >= EFX_FARCH_FILTER_TABLE_COUNT)
+ return -ENOENT;
+ table = &state->table[table_id];
+
+ filter_idx = efx_farch_filter_id_index(filter_id);
+ if (filter_idx >= table->size)
+ return -ENOENT;
+ spec = &table->spec[filter_idx];
+
+ spin_lock_bh(&efx->filter_lock);
+
+ if (test_bit(filter_idx, table->used_bitmap) &&
+ spec->priority == priority) {
+ efx_farch_filter_table_clear_entry(efx, table, filter_idx);
+ rc = 0;
+ } else {
+ rc = -ENOENT;
+ }
+
+ spin_unlock_bh(&efx->filter_lock);
+
+ return rc;
+}
+
+int efx_farch_filter_get_safe(struct efx_nic *efx,
+ enum efx_filter_priority priority,
+ u32 filter_id, struct efx_filter_spec *spec_buf)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ enum efx_farch_filter_table_id table_id;
+ struct efx_farch_filter_table *table;
+ struct efx_farch_filter_spec *spec;
+ unsigned int filter_idx;
+ int rc;
+
+ table_id = efx_farch_filter_id_table_id(filter_id);
+ if ((unsigned int)table_id >= EFX_FARCH_FILTER_TABLE_COUNT)
+ return -ENOENT;
+ table = &state->table[table_id];
+
+ filter_idx = efx_farch_filter_id_index(filter_id);
+ if (filter_idx >= table->size)
+ return -ENOENT;
+ spec = &table->spec[filter_idx];
+
+ spin_lock_bh(&efx->filter_lock);
+
+ if (test_bit(filter_idx, table->used_bitmap) &&
+ spec->priority == priority) {
+ efx_farch_filter_to_gen_spec(spec_buf, spec);
+ rc = 0;
+ } else {
+ rc = -ENOENT;
+ }
+
+ spin_unlock_bh(&efx->filter_lock);
+
+ return rc;
+}
+
+static void
+efx_farch_filter_table_clear(struct efx_nic *efx,
+ enum efx_farch_filter_table_id table_id,
+ enum efx_filter_priority priority)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_table *table = &state->table[table_id];
+ unsigned int filter_idx;
+
+ spin_lock_bh(&efx->filter_lock);
+ for (filter_idx = 0; filter_idx < table->size; ++filter_idx)
+ if (table->spec[filter_idx].priority <= priority)
+ efx_farch_filter_table_clear_entry(efx, table,
+ filter_idx);
+ spin_unlock_bh(&efx->filter_lock);
+}
+
+void efx_farch_filter_clear_rx(struct efx_nic *efx,
+ enum efx_filter_priority priority)
+{
+ efx_farch_filter_table_clear(efx, EFX_FARCH_FILTER_TABLE_RX_IP,
+ priority);
+ efx_farch_filter_table_clear(efx, EFX_FARCH_FILTER_TABLE_RX_MAC,
+ priority);
+}
+
+u32 efx_farch_filter_count_rx_used(struct efx_nic *efx,
+ enum efx_filter_priority priority)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ enum efx_farch_filter_table_id table_id;
+ struct efx_farch_filter_table *table;
+ unsigned int filter_idx;
+ u32 count = 0;
+
+ spin_lock_bh(&efx->filter_lock);
+
+ for (table_id = EFX_FARCH_FILTER_TABLE_RX_IP;
+ table_id <= EFX_FARCH_FILTER_TABLE_RX_DEF;
+ table_id++) {
+ table = &state->table[table_id];
+ for (filter_idx = 0; filter_idx < table->size; filter_idx++) {
+ if (test_bit(filter_idx, table->used_bitmap) &&
+ table->spec[filter_idx].priority == priority)
+ ++count;
+ }
+ }
+
+ spin_unlock_bh(&efx->filter_lock);
+
+ return count;
+}
+
+s32 efx_farch_filter_get_rx_ids(struct efx_nic *efx,
+ enum efx_filter_priority priority,
+ u32 *buf, u32 size)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ enum efx_farch_filter_table_id table_id;
+ struct efx_farch_filter_table *table;
+ unsigned int filter_idx;
+ s32 count = 0;
+
+ spin_lock_bh(&efx->filter_lock);
+
+ for (table_id = EFX_FARCH_FILTER_TABLE_RX_IP;
+ table_id <= EFX_FARCH_FILTER_TABLE_RX_DEF;
+ table_id++) {
+ table = &state->table[table_id];
+ for (filter_idx = 0; filter_idx < table->size; filter_idx++) {
+ if (test_bit(filter_idx, table->used_bitmap) &&
+ table->spec[filter_idx].priority == priority) {
+ if (count == size) {
+ count = -EMSGSIZE;
+ goto out;
+ }
+ buf[count++] = efx_farch_filter_make_id(
+ &table->spec[filter_idx], filter_idx);
+ }
+ }
+ }
+out:
+ spin_unlock_bh(&efx->filter_lock);
+
+ return count;
+}
+
+/* Restore filter stater after reset */
+void efx_farch_filter_table_restore(struct efx_nic *efx)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ enum efx_farch_filter_table_id table_id;
+ struct efx_farch_filter_table *table;
+ efx_oword_t filter;
+ unsigned int filter_idx;
+
+ spin_lock_bh(&efx->filter_lock);
+
+ for (table_id = 0; table_id < EFX_FARCH_FILTER_TABLE_COUNT; table_id++) {
+ table = &state->table[table_id];
+
+ /* Check whether this is a regular register table */
+ if (table->step == 0)
+ continue;
+
+ for (filter_idx = 0; filter_idx < table->size; filter_idx++) {
+ if (!test_bit(filter_idx, table->used_bitmap))
+ continue;
+ efx_farch_filter_build(&filter, &table->spec[filter_idx]);
+ efx_writeo(efx, &filter,
+ table->offset + table->step * filter_idx);
+ }
+ }
+
+ efx_farch_filter_push_rx_config(efx);
+ efx_farch_filter_push_tx_limits(efx);
+
+ spin_unlock_bh(&efx->filter_lock);
+}
+
+void efx_farch_filter_table_remove(struct efx_nic *efx)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ enum efx_farch_filter_table_id table_id;
+
+ for (table_id = 0; table_id < EFX_FARCH_FILTER_TABLE_COUNT; table_id++) {
+ kfree(state->table[table_id].used_bitmap);
+ vfree(state->table[table_id].spec);
+ }
+ kfree(state);
+}
+
+int efx_farch_filter_table_probe(struct efx_nic *efx)
+{
+ struct efx_farch_filter_state *state;
+ struct efx_farch_filter_table *table;
+ unsigned table_id;
+
+ state = kzalloc(sizeof(struct efx_farch_filter_state), GFP_KERNEL);
+ if (!state)
+ return -ENOMEM;
+ efx->filter_state = state;
+
+ if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0) {
+ table = &state->table[EFX_FARCH_FILTER_TABLE_RX_IP];
+ table->id = EFX_FARCH_FILTER_TABLE_RX_IP;
+ table->offset = FR_BZ_RX_FILTER_TBL0;
+ table->size = FR_BZ_RX_FILTER_TBL0_ROWS;
+ table->step = FR_BZ_RX_FILTER_TBL0_STEP;
+ }
+
+ if (efx_nic_rev(efx) >= EFX_REV_SIENA_A0) {
+ table = &state->table[EFX_FARCH_FILTER_TABLE_RX_MAC];
+ table->id = EFX_FARCH_FILTER_TABLE_RX_MAC;
+ table->offset = FR_CZ_RX_MAC_FILTER_TBL0;
+ table->size = FR_CZ_RX_MAC_FILTER_TBL0_ROWS;
+ table->step = FR_CZ_RX_MAC_FILTER_TBL0_STEP;
+
+ table = &state->table[EFX_FARCH_FILTER_TABLE_RX_DEF];
+ table->id = EFX_FARCH_FILTER_TABLE_RX_DEF;
+ table->size = EFX_FARCH_FILTER_SIZE_RX_DEF;
+
+ table = &state->table[EFX_FARCH_FILTER_TABLE_TX_MAC];
+ table->id = EFX_FARCH_FILTER_TABLE_TX_MAC;
+ table->offset = FR_CZ_TX_MAC_FILTER_TBL0;
+ table->size = FR_CZ_TX_MAC_FILTER_TBL0_ROWS;
+ table->step = FR_CZ_TX_MAC_FILTER_TBL0_STEP;
+ }
+
+ for (table_id = 0; table_id < EFX_FARCH_FILTER_TABLE_COUNT; table_id++) {
+ table = &state->table[table_id];
+ if (table->size == 0)
+ continue;
+ table->used_bitmap = kcalloc(BITS_TO_LONGS(table->size),
+ sizeof(unsigned long),
+ GFP_KERNEL);
+ if (!table->used_bitmap)
+ goto fail;
+ table->spec = vzalloc(table->size * sizeof(*table->spec));
+ if (!table->spec)
+ goto fail;
+ }
+
+ if (state->table[EFX_FARCH_FILTER_TABLE_RX_DEF].size) {
+ /* RX default filters must always exist */
+ unsigned i;
+ for (i = 0; i < EFX_FARCH_FILTER_SIZE_RX_DEF; i++)
+ efx_farch_filter_reset_rx_def(efx, i);
+ }
+
+ efx_farch_filter_push_rx_config(efx);
+
+ return 0;
+
+fail:
+ efx_farch_filter_table_remove(efx);
+ return -ENOMEM;
+}
+
+/* Update scatter enable flags for filters pointing to our own RX queues */
+void efx_farch_filter_update_rx_scatter(struct efx_nic *efx)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ enum efx_farch_filter_table_id table_id;
+ struct efx_farch_filter_table *table;
+ efx_oword_t filter;
+ unsigned int filter_idx;
+
+ spin_lock_bh(&efx->filter_lock);
+
+ for (table_id = EFX_FARCH_FILTER_TABLE_RX_IP;
+ table_id <= EFX_FARCH_FILTER_TABLE_RX_DEF;
+ table_id++) {
+ table = &state->table[table_id];
+
+ for (filter_idx = 0; filter_idx < table->size; filter_idx++) {
+ if (!test_bit(filter_idx, table->used_bitmap) ||
+ table->spec[filter_idx].dmaq_id >=
+ efx->n_rx_channels)
+ continue;
+
+ if (efx->rx_scatter)
+ table->spec[filter_idx].flags |=
+ EFX_FILTER_FLAG_RX_SCATTER;
+ else
+ table->spec[filter_idx].flags &=
+ ~EFX_FILTER_FLAG_RX_SCATTER;
+
+ if (table_id == EFX_FARCH_FILTER_TABLE_RX_DEF)
+ /* Pushed by efx_farch_filter_push_rx_config() */
+ continue;
+
+ efx_farch_filter_build(&filter, &table->spec[filter_idx]);
+ efx_writeo(efx, &filter,
+ table->offset + table->step * filter_idx);
+ }
+ }
+
+ efx_farch_filter_push_rx_config(efx);
+
+ spin_unlock_bh(&efx->filter_lock);
+}
+
+#ifdef CONFIG_RFS_ACCEL
+
+s32 efx_farch_filter_rfs_insert(struct efx_nic *efx,
+ struct efx_filter_spec *gen_spec)
+{
+ return efx_farch_filter_insert(efx, gen_spec, true);
+}
+
+bool efx_farch_filter_rfs_expire_one(struct efx_nic *efx, u32 flow_id,
+ unsigned int index)
+{
+ struct efx_farch_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_table *table =
+ &state->table[EFX_FARCH_FILTER_TABLE_RX_IP];
+
+ if (test_bit(index, table->used_bitmap) &&
+ table->spec[index].priority == EFX_FILTER_PRI_HINT &&
+ rps_may_expire_flow(efx->net_dev, table->spec[index].dmaq_id,
+ flow_id, index)) {
+ efx_farch_filter_table_clear_entry(efx, table, index);
+ return true;
+ }
+
+ return false;
+}
+
+#endif /* CONFIG_RFS_ACCEL */
diff --git a/drivers/net/ethernet/sfc/filter.c b/drivers/net/ethernet/sfc/filter.c
deleted file mode 100644
index ad66376..0000000
--- a/drivers/net/ethernet/sfc/filter.c
+++ /dev/null
@@ -1,1244 +0,0 @@
-/****************************************************************************
- * Driver for Solarflare Solarstorm network controllers and boards
- * Copyright 2005-2010 Solarflare Communications Inc.
- *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms of the GNU General Public License version 2 as published
- * by the Free Software Foundation, incorporated herein by reference.
- */
-
-#include <linux/in.h>
-#include <net/ip.h>
-#include "efx.h"
-#include "filter.h"
-#include "io.h"
-#include "nic.h"
-#include "farch_regs.h"
-
-/* "Fudge factors" - difference between programmed value and actual depth.
- * Due to pipelined implementation we need to program H/W with a value that
- * is larger than the hop limit we want.
- */
-#define EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD 3
-#define EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL 1
-
-/* Hard maximum search limit. Hardware will time-out beyond 200-something.
- * We also need to avoid infinite loops in efx_farch_filter_search() when the
- * table is full.
- */
-#define EFX_FARCH_FILTER_CTL_SRCH_MAX 200
-
-/* Don't try very hard to find space for performance hints, as this is
- * counter-productive. */
-#define EFX_FARCH_FILTER_CTL_SRCH_HINT_MAX 5
-
-enum efx_farch_filter_type {
- EFX_FARCH_FILTER_TCP_FULL = 0,
- EFX_FARCH_FILTER_TCP_WILD,
- EFX_FARCH_FILTER_UDP_FULL,
- EFX_FARCH_FILTER_UDP_WILD,
- EFX_FARCH_FILTER_MAC_FULL = 4,
- EFX_FARCH_FILTER_MAC_WILD,
- EFX_FARCH_FILTER_UC_DEF = 8,
- EFX_FARCH_FILTER_MC_DEF,
- EFX_FARCH_FILTER_TYPE_COUNT, /* number of specific types */
-};
-
-enum efx_farch_filter_table_id {
- EFX_FARCH_FILTER_TABLE_RX_IP = 0,
- EFX_FARCH_FILTER_TABLE_RX_MAC,
- EFX_FARCH_FILTER_TABLE_RX_DEF,
- EFX_FARCH_FILTER_TABLE_TX_MAC,
- EFX_FARCH_FILTER_TABLE_COUNT,
-};
-
-enum efx_farch_filter_index {
- EFX_FARCH_FILTER_INDEX_UC_DEF,
- EFX_FARCH_FILTER_INDEX_MC_DEF,
- EFX_FARCH_FILTER_SIZE_RX_DEF,
-};
-
-struct efx_farch_filter_spec {
- u8 type:4;
- u8 priority:4;
- u8 flags;
- u16 dmaq_id;
- u32 data[3];
-};
-
-struct efx_farch_filter_table {
- enum efx_farch_filter_table_id id;
- u32 offset; /* address of table relative to BAR */
- unsigned size; /* number of entries */
- unsigned step; /* step between entries */
- unsigned used; /* number currently used */
- unsigned long *used_bitmap;
- struct efx_farch_filter_spec *spec;
- unsigned search_limit[EFX_FARCH_FILTER_TYPE_COUNT];
-};
-
-struct efx_farch_filter_state {
- struct efx_farch_filter_table table[EFX_FARCH_FILTER_TABLE_COUNT];
-};
-
-static void
-efx_farch_filter_table_clear_entry(struct efx_nic *efx,
- struct efx_farch_filter_table *table,
- unsigned int filter_idx);
-
-/* The filter hash function is LFSR polynomial x^16 + x^3 + 1 of a 32-bit
- * key derived from the n-tuple. The initial LFSR state is 0xffff. */
-static u16 efx_farch_filter_hash(u32 key)
-{
- u16 tmp;
-
- /* First 16 rounds */
- tmp = 0x1fff ^ key >> 16;
- tmp = tmp ^ tmp >> 3 ^ tmp >> 6;
- tmp = tmp ^ tmp >> 9;
- /* Last 16 rounds */
- tmp = tmp ^ tmp << 13 ^ key;
- tmp = tmp ^ tmp >> 3 ^ tmp >> 6;
- return tmp ^ tmp >> 9;
-}
-
-/* To allow for hash collisions, filter search continues at these
- * increments from the first possible entry selected by the hash. */
-static u16 efx_farch_filter_increment(u32 key)
-{
- return key * 2 - 1;
-}
-
-static enum efx_farch_filter_table_id
-efx_farch_filter_spec_table_id(const struct efx_farch_filter_spec *spec)
-{
- BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
- (EFX_FARCH_FILTER_TCP_FULL >> 2));
- BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
- (EFX_FARCH_FILTER_TCP_WILD >> 2));
- BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
- (EFX_FARCH_FILTER_UDP_FULL >> 2));
- BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
- (EFX_FARCH_FILTER_UDP_WILD >> 2));
- BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_MAC !=
- (EFX_FARCH_FILTER_MAC_FULL >> 2));
- BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_MAC !=
- (EFX_FARCH_FILTER_MAC_WILD >> 2));
- BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_TX_MAC !=
- EFX_FARCH_FILTER_TABLE_RX_MAC + 2);
- return (spec->type >> 2) + ((spec->flags & EFX_FILTER_FLAG_TX) ? 2 : 0);
-}
-
-static void efx_farch_filter_push_rx_config(struct efx_nic *efx)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- struct efx_farch_filter_table *table;
- efx_oword_t filter_ctl;
-
- efx_reado(efx, &filter_ctl, FR_BZ_RX_FILTER_CTL);
-
- table = &state->table[EFX_FARCH_FILTER_TABLE_RX_IP];
- EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_TCP_FULL_SRCH_LIMIT,
- table->search_limit[EFX_FARCH_FILTER_TCP_FULL] +
- EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
- EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_TCP_WILD_SRCH_LIMIT,
- table->search_limit[EFX_FARCH_FILTER_TCP_WILD] +
- EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
- EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_UDP_FULL_SRCH_LIMIT,
- table->search_limit[EFX_FARCH_FILTER_UDP_FULL] +
- EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
- EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_UDP_WILD_SRCH_LIMIT,
- table->search_limit[EFX_FARCH_FILTER_UDP_WILD] +
- EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
-
- table = &state->table[EFX_FARCH_FILTER_TABLE_RX_MAC];
- if (table->size) {
- EFX_SET_OWORD_FIELD(
- filter_ctl, FRF_CZ_ETHERNET_FULL_SEARCH_LIMIT,
- table->search_limit[EFX_FARCH_FILTER_MAC_FULL] +
- EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
- EFX_SET_OWORD_FIELD(
- filter_ctl, FRF_CZ_ETHERNET_WILDCARD_SEARCH_LIMIT,
- table->search_limit[EFX_FARCH_FILTER_MAC_WILD] +
- EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
- }
-
- table = &state->table[EFX_FARCH_FILTER_TABLE_RX_DEF];
- if (table->size) {
- EFX_SET_OWORD_FIELD(
- filter_ctl, FRF_CZ_UNICAST_NOMATCH_Q_ID,
- table->spec[EFX_FARCH_FILTER_INDEX_UC_DEF].dmaq_id);
- EFX_SET_OWORD_FIELD(
- filter_ctl, FRF_CZ_UNICAST_NOMATCH_RSS_ENABLED,
- !!(table->spec[EFX_FARCH_FILTER_INDEX_UC_DEF].flags &
- EFX_FILTER_FLAG_RX_RSS));
- EFX_SET_OWORD_FIELD(
- filter_ctl, FRF_CZ_MULTICAST_NOMATCH_Q_ID,
- table->spec[EFX_FARCH_FILTER_INDEX_MC_DEF].dmaq_id);
- EFX_SET_OWORD_FIELD(
- filter_ctl, FRF_CZ_MULTICAST_NOMATCH_RSS_ENABLED,
- !!(table->spec[EFX_FARCH_FILTER_INDEX_MC_DEF].flags &
- EFX_FILTER_FLAG_RX_RSS));
-
- /* There is a single bit to enable RX scatter for all
- * unmatched packets. Only set it if scatter is
- * enabled in both filter specs.
- */
- EFX_SET_OWORD_FIELD(
- filter_ctl, FRF_BZ_SCATTER_ENBL_NO_MATCH_Q,
- !!(table->spec[EFX_FARCH_FILTER_INDEX_UC_DEF].flags &
- table->spec[EFX_FARCH_FILTER_INDEX_MC_DEF].flags &
- EFX_FILTER_FLAG_RX_SCATTER));
- } else if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0) {
- /* We don't expose 'default' filters because unmatched
- * packets always go to the queue number found in the
- * RSS table. But we still need to set the RX scatter
- * bit here.
- */
- EFX_SET_OWORD_FIELD(
- filter_ctl, FRF_BZ_SCATTER_ENBL_NO_MATCH_Q,
- efx->rx_scatter);
- }
-
- efx_writeo(efx, &filter_ctl, FR_BZ_RX_FILTER_CTL);
-}
-
-static void efx_farch_filter_push_tx_limits(struct efx_nic *efx)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- struct efx_farch_filter_table *table;
- efx_oword_t tx_cfg;
-
- efx_reado(efx, &tx_cfg, FR_AZ_TX_CFG);
-
- table = &state->table[EFX_FARCH_FILTER_TABLE_TX_MAC];
- if (table->size) {
- EFX_SET_OWORD_FIELD(
- tx_cfg, FRF_CZ_TX_ETH_FILTER_FULL_SEARCH_RANGE,
- table->search_limit[EFX_FARCH_FILTER_MAC_FULL] +
- EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
- EFX_SET_OWORD_FIELD(
- tx_cfg, FRF_CZ_TX_ETH_FILTER_WILD_SEARCH_RANGE,
- table->search_limit[EFX_FARCH_FILTER_MAC_WILD] +
- EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
- }
-
- efx_writeo(efx, &tx_cfg, FR_AZ_TX_CFG);
-}
-
-static int
-efx_farch_filter_from_gen_spec(struct efx_farch_filter_spec *spec,
- const struct efx_filter_spec *gen_spec)
-{
- bool is_full = false;
-
- if ((gen_spec->flags & EFX_FILTER_FLAG_RX_RSS) &&
- gen_spec->rss_context != EFX_FILTER_RSS_CONTEXT_DEFAULT)
- return -EINVAL;
-
- spec->priority = gen_spec->priority;
- spec->flags = gen_spec->flags;
- spec->dmaq_id = gen_spec->dmaq_id;
-
- switch (gen_spec->match_flags) {
- case (EFX_FILTER_MATCH_ETHER_TYPE | EFX_FILTER_MATCH_IP_PROTO |
- EFX_FILTER_MATCH_LOC_HOST | EFX_FILTER_MATCH_LOC_PORT |
- EFX_FILTER_MATCH_REM_HOST | EFX_FILTER_MATCH_REM_PORT):
- is_full = true;
- /* fall through */
- case (EFX_FILTER_MATCH_ETHER_TYPE | EFX_FILTER_MATCH_IP_PROTO |
- EFX_FILTER_MATCH_LOC_HOST | EFX_FILTER_MATCH_LOC_PORT): {
- __be32 rhost, host1, host2;
- __be16 rport, port1, port2;
-
- EFX_BUG_ON_PARANOID(!(gen_spec->flags & EFX_FILTER_FLAG_RX));
-
- if (gen_spec->ether_type != htons(ETH_P_IP))
- return -EPROTONOSUPPORT;
- if (gen_spec->loc_port == 0 ||
- (is_full && gen_spec->rem_port == 0))
- return -EADDRNOTAVAIL;
- switch (gen_spec->ip_proto) {
- case IPPROTO_TCP:
- spec->type = (is_full ? EFX_FARCH_FILTER_TCP_FULL :
- EFX_FARCH_FILTER_TCP_WILD);
- break;
- case IPPROTO_UDP:
- spec->type = (is_full ? EFX_FARCH_FILTER_UDP_FULL :
- EFX_FARCH_FILTER_UDP_WILD);
- break;
- default:
- return -EPROTONOSUPPORT;
- }
-
- /* Filter is constructed in terms of source and destination,
- * with the odd wrinkle that the ports are swapped in a UDP
- * wildcard filter. We need to convert from local and remote
- * (= zero for wildcard) addresses.
- */
- rhost = is_full ? gen_spec->rem_host[0] : 0;
- rport = is_full ? gen_spec->rem_port : 0;
- host1 = rhost;
- host2 = gen_spec->loc_host[0];
- if (!is_full && gen_spec->ip_proto == IPPROTO_UDP) {
- port1 = gen_spec->loc_port;
- port2 = rport;
- } else {
- port1 = rport;
- port2 = gen_spec->loc_port;
- }
- spec->data[0] = ntohl(host1) << 16 | ntohs(port1);
- spec->data[1] = ntohs(port2) << 16 | ntohl(host1) >> 16;
- spec->data[2] = ntohl(host2);
-
- break;
- }
-
- case EFX_FILTER_MATCH_LOC_MAC | EFX_FILTER_MATCH_OUTER_VID:
- is_full = true;
- /* fall through */
- case EFX_FILTER_MATCH_LOC_MAC:
- spec->type = (is_full ? EFX_FARCH_FILTER_MAC_FULL :
- EFX_FARCH_FILTER_MAC_WILD);
- spec->data[0] = is_full ? ntohs(gen_spec->outer_vid) : 0;
- spec->data[1] = (gen_spec->loc_mac[2] << 24 |
- gen_spec->loc_mac[3] << 16 |
- gen_spec->loc_mac[4] << 8 |
- gen_spec->loc_mac[5]);
- spec->data[2] = (gen_spec->loc_mac[0] << 8 |
- gen_spec->loc_mac[1]);
- break;
-
- case EFX_FILTER_MATCH_LOC_MAC_IG:
- spec->type = (is_multicast_ether_addr(gen_spec->loc_mac) ?
- EFX_FARCH_FILTER_MC_DEF :
- EFX_FARCH_FILTER_UC_DEF);
- memset(spec->data, 0, sizeof(spec->data)); /* ensure equality */
- break;
-
- default:
- return -EPROTONOSUPPORT;
- }
-
- return 0;
-}
-
-static void
-efx_farch_filter_to_gen_spec(struct efx_filter_spec *gen_spec,
- const struct efx_farch_filter_spec *spec)
-{
- bool is_full = false;
-
- /* *gen_spec should be completely initialised, to be consistent
- * with efx_filter_init_{rx,tx}() and in case we want to copy
- * it back to userland.
- */
- memset(gen_spec, 0, sizeof(*gen_spec));
-
- gen_spec->priority = spec->priority;
- gen_spec->flags = spec->flags;
- gen_spec->dmaq_id = spec->dmaq_id;
-
- switch (spec->type) {
- case EFX_FARCH_FILTER_TCP_FULL:
- case EFX_FARCH_FILTER_UDP_FULL:
- is_full = true;
- /* fall through */
- case EFX_FARCH_FILTER_TCP_WILD:
- case EFX_FARCH_FILTER_UDP_WILD: {
- __be32 host1, host2;
- __be16 port1, port2;
-
- gen_spec->match_flags =
- EFX_FILTER_MATCH_ETHER_TYPE |
- EFX_FILTER_MATCH_IP_PROTO |
- EFX_FILTER_MATCH_LOC_HOST | EFX_FILTER_MATCH_LOC_PORT;
- if (is_full)
- gen_spec->match_flags |= (EFX_FILTER_MATCH_REM_HOST |
- EFX_FILTER_MATCH_REM_PORT);
- gen_spec->ether_type = htons(ETH_P_IP);
- gen_spec->ip_proto =
- (spec->type == EFX_FARCH_FILTER_TCP_FULL ||
- spec->type == EFX_FARCH_FILTER_TCP_WILD) ?
- IPPROTO_TCP : IPPROTO_UDP;
-
- host1 = htonl(spec->data[0] >> 16 | spec->data[1] << 16);
- port1 = htons(spec->data[0]);
- host2 = htonl(spec->data[2]);
- port2 = htons(spec->data[1] >> 16);
- if (spec->flags & EFX_FILTER_FLAG_TX) {
- gen_spec->loc_host[0] = host1;
- gen_spec->rem_host[0] = host2;
- } else {
- gen_spec->loc_host[0] = host2;
- gen_spec->rem_host[0] = host1;
- }
- if (!!(gen_spec->flags & EFX_FILTER_FLAG_TX) ^
- (!is_full && gen_spec->ip_proto == IPPROTO_UDP)) {
- gen_spec->loc_port = port1;
- gen_spec->rem_port = port2;
- } else {
- gen_spec->loc_port = port2;
- gen_spec->rem_port = port1;
- }
-
- break;
- }
-
- case EFX_FARCH_FILTER_MAC_FULL:
- is_full = true;
- /* fall through */
- case EFX_FARCH_FILTER_MAC_WILD:
- gen_spec->match_flags = EFX_FILTER_MATCH_LOC_MAC;
- if (is_full)
- gen_spec->match_flags |= EFX_FILTER_MATCH_OUTER_VID;
- gen_spec->loc_mac[0] = spec->data[2] >> 8;
- gen_spec->loc_mac[1] = spec->data[2];
- gen_spec->loc_mac[2] = spec->data[1] >> 24;
- gen_spec->loc_mac[3] = spec->data[1] >> 16;
- gen_spec->loc_mac[4] = spec->data[1] >> 8;
- gen_spec->loc_mac[5] = spec->data[1];
- gen_spec->outer_vid = htons(spec->data[0]);
- break;
-
- case EFX_FARCH_FILTER_UC_DEF:
- case EFX_FARCH_FILTER_MC_DEF:
- gen_spec->match_flags = EFX_FILTER_MATCH_LOC_MAC_IG;
- gen_spec->loc_mac[0] = spec->type == EFX_FARCH_FILTER_MC_DEF;
- break;
-
- default:
- WARN_ON(1);
- break;
- }
-}
-
-static void
-efx_farch_filter_reset_rx_def(struct efx_nic *efx, unsigned filter_idx)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- struct efx_farch_filter_table *table =
- &state->table[EFX_FARCH_FILTER_TABLE_RX_DEF];
- struct efx_farch_filter_spec *spec = &table->spec[filter_idx];
-
- /* If there's only one channel then disable RSS for non VF
- * traffic, thereby allowing VFs to use RSS when the PF can't.
- */
- spec->type = EFX_FARCH_FILTER_UC_DEF + filter_idx;
- spec->priority = EFX_FILTER_PRI_MANUAL;
- spec->flags = (EFX_FILTER_FLAG_RX |
- (efx->n_rx_channels > 1 ? EFX_FILTER_FLAG_RX_RSS : 0) |
- (efx->rx_scatter ? EFX_FILTER_FLAG_RX_SCATTER : 0));
- spec->dmaq_id = 0;
- table->used_bitmap[0] |= 1 << filter_idx;
-}
-
-/* Build a filter entry and return its n-tuple key. */
-static u32 efx_farch_filter_build(efx_oword_t *filter,
- struct efx_farch_filter_spec *spec)
-{
- u32 data3;
-
- switch (efx_farch_filter_spec_table_id(spec)) {
- case EFX_FARCH_FILTER_TABLE_RX_IP: {
- bool is_udp = (spec->type == EFX_FARCH_FILTER_UDP_FULL ||
- spec->type == EFX_FARCH_FILTER_UDP_WILD);
- EFX_POPULATE_OWORD_7(
- *filter,
- FRF_BZ_RSS_EN,
- !!(spec->flags & EFX_FILTER_FLAG_RX_RSS),
- FRF_BZ_SCATTER_EN,
- !!(spec->flags & EFX_FILTER_FLAG_RX_SCATTER),
- FRF_BZ_TCP_UDP, is_udp,
- FRF_BZ_RXQ_ID, spec->dmaq_id,
- EFX_DWORD_2, spec->data[2],
- EFX_DWORD_1, spec->data[1],
- EFX_DWORD_0, spec->data[0]);
- data3 = is_udp;
- break;
- }
-
- case EFX_FARCH_FILTER_TABLE_RX_MAC: {
- bool is_wild = spec->type == EFX_FARCH_FILTER_MAC_WILD;
- EFX_POPULATE_OWORD_7(
- *filter,
- FRF_CZ_RMFT_RSS_EN,
- !!(spec->flags & EFX_FILTER_FLAG_RX_RSS),
- FRF_CZ_RMFT_SCATTER_EN,
- !!(spec->flags & EFX_FILTER_FLAG_RX_SCATTER),
- FRF_CZ_RMFT_RXQ_ID, spec->dmaq_id,
- FRF_CZ_RMFT_WILDCARD_MATCH, is_wild,
- FRF_CZ_RMFT_DEST_MAC_HI, spec->data[2],
- FRF_CZ_RMFT_DEST_MAC_LO, spec->data[1],
- FRF_CZ_RMFT_VLAN_ID, spec->data[0]);
- data3 = is_wild;
- break;
- }
-
- case EFX_FARCH_FILTER_TABLE_TX_MAC: {
- bool is_wild = spec->type == EFX_FARCH_FILTER_MAC_WILD;
- EFX_POPULATE_OWORD_5(*filter,
- FRF_CZ_TMFT_TXQ_ID, spec->dmaq_id,
- FRF_CZ_TMFT_WILDCARD_MATCH, is_wild,
- FRF_CZ_TMFT_SRC_MAC_HI, spec->data[2],
- FRF_CZ_TMFT_SRC_MAC_LO, spec->data[1],
- FRF_CZ_TMFT_VLAN_ID, spec->data[0]);
- data3 = is_wild | spec->dmaq_id << 1;
- break;
- }
-
- default:
- BUG();
- }
-
- return spec->data[0] ^ spec->data[1] ^ spec->data[2] ^ data3;
-}
-
-static bool efx_farch_filter_equal(const struct efx_farch_filter_spec *left,
- const struct efx_farch_filter_spec *right)
-{
- if (left->type != right->type ||
- memcmp(left->data, right->data, sizeof(left->data)))
- return false;
-
- if (left->flags & EFX_FILTER_FLAG_TX &&
- left->dmaq_id != right->dmaq_id)
- return false;
-
- return true;
-}
-
-/*
- * Construct/deconstruct external filter IDs. At least the RX filter
- * IDs must be ordered by matching priority, for RX NFC semantics.
- *
- * Deconstruction needs to be robust against invalid IDs so that
- * efx_filter_remove_id_safe() and efx_filter_get_filter_safe() can
- * accept user-provided IDs.
- */
-
-#define EFX_FARCH_FILTER_MATCH_PRI_COUNT 5
-
-static const u8 efx_farch_filter_type_match_pri[EFX_FARCH_FILTER_TYPE_COUNT] = {
- [EFX_FARCH_FILTER_TCP_FULL] = 0,
- [EFX_FARCH_FILTER_UDP_FULL] = 0,
- [EFX_FARCH_FILTER_TCP_WILD] = 1,
- [EFX_FARCH_FILTER_UDP_WILD] = 1,
- [EFX_FARCH_FILTER_MAC_FULL] = 2,
- [EFX_FARCH_FILTER_MAC_WILD] = 3,
- [EFX_FARCH_FILTER_UC_DEF] = 4,
- [EFX_FARCH_FILTER_MC_DEF] = 4,
-};
-
-static const enum efx_farch_filter_table_id efx_farch_filter_range_table[] = {
- EFX_FARCH_FILTER_TABLE_RX_IP, /* RX match pri 0 */
- EFX_FARCH_FILTER_TABLE_RX_IP,
- EFX_FARCH_FILTER_TABLE_RX_MAC,
- EFX_FARCH_FILTER_TABLE_RX_MAC,
- EFX_FARCH_FILTER_TABLE_RX_DEF, /* RX match pri 4 */
- EFX_FARCH_FILTER_TABLE_TX_MAC, /* TX match pri 0 */
- EFX_FARCH_FILTER_TABLE_TX_MAC, /* TX match pri 1 */
-};
-
-#define EFX_FARCH_FILTER_INDEX_WIDTH 13
-#define EFX_FARCH_FILTER_INDEX_MASK ((1 << EFX_FARCH_FILTER_INDEX_WIDTH) - 1)
-
-static inline u32
-efx_farch_filter_make_id(const struct efx_farch_filter_spec *spec,
- unsigned int index)
-{
- unsigned int range;
-
- range = efx_farch_filter_type_match_pri[spec->type];
- if (!(spec->flags & EFX_FILTER_FLAG_RX))
- range += EFX_FARCH_FILTER_MATCH_PRI_COUNT;
-
- return range << EFX_FARCH_FILTER_INDEX_WIDTH | index;
-}
-
-static inline enum efx_farch_filter_table_id
-efx_farch_filter_id_table_id(u32 id)
-{
- unsigned int range = id >> EFX_FARCH_FILTER_INDEX_WIDTH;
-
- if (range < ARRAY_SIZE(efx_farch_filter_range_table))
- return efx_farch_filter_range_table[range];
- else
- return EFX_FARCH_FILTER_TABLE_COUNT; /* invalid */
-}
-
-static inline unsigned int efx_farch_filter_id_index(u32 id)
-{
- return id & EFX_FARCH_FILTER_INDEX_MASK;
-}
-
-u32 efx_filter_get_rx_id_limit(struct efx_nic *efx)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- unsigned int range = EFX_FARCH_FILTER_MATCH_PRI_COUNT - 1;
- enum efx_farch_filter_table_id table_id;
-
- do {
- table_id = efx_farch_filter_range_table[range];
- if (state->table[table_id].size != 0)
- return range << EFX_FARCH_FILTER_INDEX_WIDTH |
- state->table[table_id].size;
- } while (range--);
-
- return 0;
-}
-
-/**
- * efx_filter_insert_filter - add or replace a filter
- * @efx: NIC in which to insert the filter
- * @spec: Specification for the filter
- * @replace_equal: Flag for whether the specified filter may replace an
- * existing filter with equal priority
- *
- * On success, return the filter ID.
- * On failure, return a negative error code.
- *
- * If an existing filter has equal match values to the new filter
- * spec, then the new filter might replace it, depending on the
- * relative priorities. If the existing filter has lower priority, or
- * if @replace_equal is set and it has equal priority, then it is
- * replaced. Otherwise the function fails, returning -%EPERM if
- * the existing filter has higher priority or -%EEXIST if it has
- * equal priority.
- */
-s32 efx_filter_insert_filter(struct efx_nic *efx,
- struct efx_filter_spec *gen_spec,
- bool replace_equal)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- struct efx_farch_filter_table *table;
- struct efx_farch_filter_spec spec;
- efx_oword_t filter;
- int rep_index, ins_index;
- unsigned int depth = 0;
- int rc;
-
- rc = efx_farch_filter_from_gen_spec(&spec, gen_spec);
- if (rc)
- return rc;
-
- table = &state->table[efx_farch_filter_spec_table_id(&spec)];
- if (table->size == 0)
- return -EINVAL;
-
- netif_vdbg(efx, hw, efx->net_dev,
- "%s: type %d search_limit=%d", __func__, spec.type,
- table->search_limit[spec.type]);
-
- if (table->id == EFX_FARCH_FILTER_TABLE_RX_DEF) {
- /* One filter spec per type */
- BUILD_BUG_ON(EFX_FARCH_FILTER_INDEX_UC_DEF != 0);
- BUILD_BUG_ON(EFX_FARCH_FILTER_INDEX_MC_DEF !=
- EFX_FARCH_FILTER_MC_DEF - EFX_FARCH_FILTER_UC_DEF);
- rep_index = spec.type - EFX_FARCH_FILTER_UC_DEF;
- ins_index = rep_index;
-
- spin_lock_bh(&efx->filter_lock);
- } else {
- /* Search concurrently for
- * (1) a filter to be replaced (rep_index): any filter
- * with the same match values, up to the current
- * search depth for this type, and
- * (2) the insertion point (ins_index): (1) or any
- * free slot before it or up to the maximum search
- * depth for this priority
- * We fail if we cannot find (2).
- *
- * We can stop once either
- * (a) we find (1), in which case we have definitely
- * found (2) as well; or
- * (b) we have searched exhaustively for (1), and have
- * either found (2) or searched exhaustively for it
- */
- u32 key = efx_farch_filter_build(&filter, &spec);
- unsigned int hash = efx_farch_filter_hash(key);
- unsigned int incr = efx_farch_filter_increment(key);
- unsigned int max_rep_depth = table->search_limit[spec.type];
- unsigned int max_ins_depth =
- spec.priority <= EFX_FILTER_PRI_HINT ?
- EFX_FARCH_FILTER_CTL_SRCH_HINT_MAX :
- EFX_FARCH_FILTER_CTL_SRCH_MAX;
- unsigned int i = hash & (table->size - 1);
-
- ins_index = -1;
- depth = 1;
-
- spin_lock_bh(&efx->filter_lock);
-
- for (;;) {
- if (!test_bit(i, table->used_bitmap)) {
- if (ins_index < 0)
- ins_index = i;
- } else if (efx_farch_filter_equal(&spec,
- &table->spec[i])) {
- /* Case (a) */
- if (ins_index < 0)
- ins_index = i;
- rep_index = i;
- break;
- }
-
- if (depth >= max_rep_depth &&
- (ins_index >= 0 || depth >= max_ins_depth)) {
- /* Case (b) */
- if (ins_index < 0) {
- rc = -EBUSY;
- goto out;
- }
- rep_index = -1;
- break;
- }
-
- i = (i + incr) & (table->size - 1);
- ++depth;
- }
- }
-
- /* If we found a filter to be replaced, check whether we
- * should do so
- */
- if (rep_index >= 0) {
- struct efx_farch_filter_spec *saved_spec =
- &table->spec[rep_index];
-
- if (spec.priority == saved_spec->priority && !replace_equal) {
- rc = -EEXIST;
- goto out;
- }
- if (spec.priority < saved_spec->priority) {
- rc = -EPERM;
- goto out;
- }
- }
-
- /* Insert the filter */
- if (ins_index != rep_index) {
- __set_bit(ins_index, table->used_bitmap);
- ++table->used;
- }
- table->spec[ins_index] = spec;
-
- if (table->id == EFX_FARCH_FILTER_TABLE_RX_DEF) {
- efx_farch_filter_push_rx_config(efx);
- } else {
- if (table->search_limit[spec.type] < depth) {
- table->search_limit[spec.type] = depth;
- if (spec.flags & EFX_FILTER_FLAG_TX)
- efx_farch_filter_push_tx_limits(efx);
- else
- efx_farch_filter_push_rx_config(efx);
- }
-
- efx_writeo(efx, &filter,
- table->offset + table->step * ins_index);
-
- /* If we were able to replace a filter by inserting
- * at a lower depth, clear the replaced filter
- */
- if (ins_index != rep_index && rep_index >= 0)
- efx_farch_filter_table_clear_entry(efx, table,
- rep_index);
- }
-
- netif_vdbg(efx, hw, efx->net_dev,
- "%s: filter type %d index %d rxq %u set",
- __func__, spec.type, ins_index, spec.dmaq_id);
- rc = efx_farch_filter_make_id(&spec, ins_index);
-
-out:
- spin_unlock_bh(&efx->filter_lock);
- return rc;
-}
-
-static void
-efx_farch_filter_table_clear_entry(struct efx_nic *efx,
- struct efx_farch_filter_table *table,
- unsigned int filter_idx)
-{
- static efx_oword_t filter;
-
- if (table->id == EFX_FARCH_FILTER_TABLE_RX_DEF) {
- /* RX default filters must always exist */
- efx_farch_filter_reset_rx_def(efx, filter_idx);
- efx_farch_filter_push_rx_config(efx);
- } else if (test_bit(filter_idx, table->used_bitmap)) {
- __clear_bit(filter_idx, table->used_bitmap);
- --table->used;
- memset(&table->spec[filter_idx], 0, sizeof(table->spec[0]));
-
- efx_writeo(efx, &filter,
- table->offset + table->step * filter_idx);
-
- /* If this filter required a greater search depth than
- * any other, the search limit for its type can now be
- * decreased. However, it is hard to determine that
- * unless the table has become completely empty - in
- * which case, all its search limits can be set to 0.
- */
- if (unlikely(table->used == 0)) {
- memset(table->search_limit, 0,
- sizeof(table->search_limit));
- if (table->id == EFX_FARCH_FILTER_TABLE_TX_MAC)
- efx_farch_filter_push_tx_limits(efx);
- else
- efx_farch_filter_push_rx_config(efx);
- }
- }
-}
-
-/**
- * efx_filter_remove_id_safe - remove a filter by ID, carefully
- * @efx: NIC from which to remove the filter
- * @priority: Priority of filter, as passed to @efx_filter_insert_filter
- * @filter_id: ID of filter, as returned by @efx_filter_insert_filter
- *
- * This function will range-check @filter_id, so it is safe to call
- * with a value passed from userland.
- */
-int efx_filter_remove_id_safe(struct efx_nic *efx,
- enum efx_filter_priority priority,
- u32 filter_id)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- enum efx_farch_filter_table_id table_id;
- struct efx_farch_filter_table *table;
- unsigned int filter_idx;
- struct efx_farch_filter_spec *spec;
- int rc;
-
- table_id = efx_farch_filter_id_table_id(filter_id);
- if ((unsigned int)table_id >= EFX_FARCH_FILTER_TABLE_COUNT)
- return -ENOENT;
- table = &state->table[table_id];
-
- filter_idx = efx_farch_filter_id_index(filter_id);
- if (filter_idx >= table->size)
- return -ENOENT;
- spec = &table->spec[filter_idx];
-
- spin_lock_bh(&efx->filter_lock);
-
- if (test_bit(filter_idx, table->used_bitmap) &&
- spec->priority == priority) {
- efx_farch_filter_table_clear_entry(efx, table, filter_idx);
- rc = 0;
- } else {
- rc = -ENOENT;
- }
-
- spin_unlock_bh(&efx->filter_lock);
-
- return rc;
-}
-
-/**
- * efx_filter_get_filter_safe - retrieve a filter by ID, carefully
- * @efx: NIC from which to remove the filter
- * @priority: Priority of filter, as passed to @efx_filter_insert_filter
- * @filter_id: ID of filter, as returned by @efx_filter_insert_filter
- * @spec: Buffer in which to store filter specification
- *
- * This function will range-check @filter_id, so it is safe to call
- * with a value passed from userland.
- */
-int efx_filter_get_filter_safe(struct efx_nic *efx,
- enum efx_filter_priority priority,
- u32 filter_id, struct efx_filter_spec *spec_buf)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- enum efx_farch_filter_table_id table_id;
- struct efx_farch_filter_table *table;
- struct efx_farch_filter_spec *spec;
- unsigned int filter_idx;
- int rc;
-
- table_id = efx_farch_filter_id_table_id(filter_id);
- if ((unsigned int)table_id >= EFX_FARCH_FILTER_TABLE_COUNT)
- return -ENOENT;
- table = &state->table[table_id];
-
- filter_idx = efx_farch_filter_id_index(filter_id);
- if (filter_idx >= table->size)
- return -ENOENT;
- spec = &table->spec[filter_idx];
-
- spin_lock_bh(&efx->filter_lock);
-
- if (test_bit(filter_idx, table->used_bitmap) &&
- spec->priority == priority) {
- efx_farch_filter_to_gen_spec(spec_buf, spec);
- rc = 0;
- } else {
- rc = -ENOENT;
- }
-
- spin_unlock_bh(&efx->filter_lock);
-
- return rc;
-}
-
-static void
-efx_farch_filter_table_clear(struct efx_nic *efx,
- enum efx_farch_filter_table_id table_id,
- enum efx_filter_priority priority)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- struct efx_farch_filter_table *table = &state->table[table_id];
- unsigned int filter_idx;
-
- spin_lock_bh(&efx->filter_lock);
- for (filter_idx = 0; filter_idx < table->size; ++filter_idx)
- if (table->spec[filter_idx].priority <= priority)
- efx_farch_filter_table_clear_entry(efx, table,
- filter_idx);
- spin_unlock_bh(&efx->filter_lock);
-}
-
-/**
- * efx_filter_clear_rx - remove RX filters by priority
- * @efx: NIC from which to remove the filters
- * @priority: Maximum priority to remove
- */
-void efx_filter_clear_rx(struct efx_nic *efx, enum efx_filter_priority priority)
-{
- efx_farch_filter_table_clear(efx, EFX_FARCH_FILTER_TABLE_RX_IP,
- priority);
- efx_farch_filter_table_clear(efx, EFX_FARCH_FILTER_TABLE_RX_MAC,
- priority);
-}
-
-u32 efx_filter_count_rx_used(struct efx_nic *efx,
- enum efx_filter_priority priority)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- enum efx_farch_filter_table_id table_id;
- struct efx_farch_filter_table *table;
- unsigned int filter_idx;
- u32 count = 0;
-
- spin_lock_bh(&efx->filter_lock);
-
- for (table_id = EFX_FARCH_FILTER_TABLE_RX_IP;
- table_id <= EFX_FARCH_FILTER_TABLE_RX_DEF;
- table_id++) {
- table = &state->table[table_id];
- for (filter_idx = 0; filter_idx < table->size; filter_idx++) {
- if (test_bit(filter_idx, table->used_bitmap) &&
- table->spec[filter_idx].priority == priority)
- ++count;
- }
- }
-
- spin_unlock_bh(&efx->filter_lock);
-
- return count;
-}
-
-s32 efx_filter_get_rx_ids(struct efx_nic *efx,
- enum efx_filter_priority priority,
- u32 *buf, u32 size)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- enum efx_farch_filter_table_id table_id;
- struct efx_farch_filter_table *table;
- unsigned int filter_idx;
- s32 count = 0;
-
- spin_lock_bh(&efx->filter_lock);
-
- for (table_id = EFX_FARCH_FILTER_TABLE_RX_IP;
- table_id <= EFX_FARCH_FILTER_TABLE_RX_DEF;
- table_id++) {
- table = &state->table[table_id];
- for (filter_idx = 0; filter_idx < table->size; filter_idx++) {
- if (test_bit(filter_idx, table->used_bitmap) &&
- table->spec[filter_idx].priority == priority) {
- if (count == size) {
- count = -EMSGSIZE;
- goto out;
- }
- buf[count++] = efx_farch_filter_make_id(
- &table->spec[filter_idx], filter_idx);
- }
- }
- }
-out:
- spin_unlock_bh(&efx->filter_lock);
-
- return count;
-}
-
-/* Restore filter stater after reset */
-void efx_restore_filters(struct efx_nic *efx)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- enum efx_farch_filter_table_id table_id;
- struct efx_farch_filter_table *table;
- efx_oword_t filter;
- unsigned int filter_idx;
-
- spin_lock_bh(&efx->filter_lock);
-
- for (table_id = 0; table_id < EFX_FARCH_FILTER_TABLE_COUNT; table_id++) {
- table = &state->table[table_id];
-
- /* Check whether this is a regular register table */
- if (table->step == 0)
- continue;
-
- for (filter_idx = 0; filter_idx < table->size; filter_idx++) {
- if (!test_bit(filter_idx, table->used_bitmap))
- continue;
- efx_farch_filter_build(&filter, &table->spec[filter_idx]);
- efx_writeo(efx, &filter,
- table->offset + table->step * filter_idx);
- }
- }
-
- efx_farch_filter_push_rx_config(efx);
- efx_farch_filter_push_tx_limits(efx);
-
- spin_unlock_bh(&efx->filter_lock);
-}
-
-int efx_probe_filters(struct efx_nic *efx)
-{
- struct efx_farch_filter_state *state;
- struct efx_farch_filter_table *table;
- unsigned table_id;
-
- state = kzalloc(sizeof(struct efx_farch_filter_state), GFP_KERNEL);
- if (!state)
- return -ENOMEM;
- efx->filter_state = state;
-
- spin_lock_init(&efx->filter_lock);
-
- if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0) {
-#ifdef CONFIG_RFS_ACCEL
- efx->rps_flow_id = kcalloc(FR_BZ_RX_FILTER_TBL0_ROWS,
- sizeof(*efx->rps_flow_id),
- GFP_KERNEL);
- if (!efx->rps_flow_id)
- goto fail;
-#endif
- table = &state->table[EFX_FARCH_FILTER_TABLE_RX_IP];
- table->id = EFX_FARCH_FILTER_TABLE_RX_IP;
- table->offset = FR_BZ_RX_FILTER_TBL0;
- table->size = FR_BZ_RX_FILTER_TBL0_ROWS;
- table->step = FR_BZ_RX_FILTER_TBL0_STEP;
- }
-
- if (efx_nic_rev(efx) >= EFX_REV_SIENA_A0) {
- table = &state->table[EFX_FARCH_FILTER_TABLE_RX_MAC];
- table->id = EFX_FARCH_FILTER_TABLE_RX_MAC;
- table->offset = FR_CZ_RX_MAC_FILTER_TBL0;
- table->size = FR_CZ_RX_MAC_FILTER_TBL0_ROWS;
- table->step = FR_CZ_RX_MAC_FILTER_TBL0_STEP;
-
- table = &state->table[EFX_FARCH_FILTER_TABLE_RX_DEF];
- table->id = EFX_FARCH_FILTER_TABLE_RX_DEF;
- table->size = EFX_FARCH_FILTER_SIZE_RX_DEF;
-
- table = &state->table[EFX_FARCH_FILTER_TABLE_TX_MAC];
- table->id = EFX_FARCH_FILTER_TABLE_TX_MAC;
- table->offset = FR_CZ_TX_MAC_FILTER_TBL0;
- table->size = FR_CZ_TX_MAC_FILTER_TBL0_ROWS;
- table->step = FR_CZ_TX_MAC_FILTER_TBL0_STEP;
- }
-
- for (table_id = 0; table_id < EFX_FARCH_FILTER_TABLE_COUNT; table_id++) {
- table = &state->table[table_id];
- if (table->size == 0)
- continue;
- table->used_bitmap = kcalloc(BITS_TO_LONGS(table->size),
- sizeof(unsigned long),
- GFP_KERNEL);
- if (!table->used_bitmap)
- goto fail;
- table->spec = vzalloc(table->size * sizeof(*table->spec));
- if (!table->spec)
- goto fail;
- }
-
- if (state->table[EFX_FARCH_FILTER_TABLE_RX_DEF].size) {
- /* RX default filters must always exist */
- unsigned i;
- for (i = 0; i < EFX_FARCH_FILTER_SIZE_RX_DEF; i++)
- efx_farch_filter_reset_rx_def(efx, i);
- }
-
- efx_farch_filter_push_rx_config(efx);
-
- return 0;
-
-fail:
- efx_remove_filters(efx);
- return -ENOMEM;
-}
-
-void efx_remove_filters(struct efx_nic *efx)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- enum efx_farch_filter_table_id table_id;
-
- for (table_id = 0; table_id < EFX_FARCH_FILTER_TABLE_COUNT; table_id++) {
- kfree(state->table[table_id].used_bitmap);
- vfree(state->table[table_id].spec);
- }
-#ifdef CONFIG_RFS_ACCEL
- kfree(efx->rps_flow_id);
-#endif
- kfree(state);
-}
-
-/* Update scatter enable flags for filters pointing to our own RX queues */
-void efx_filter_update_rx_scatter(struct efx_nic *efx)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- enum efx_farch_filter_table_id table_id;
- struct efx_farch_filter_table *table;
- efx_oword_t filter;
- unsigned int filter_idx;
-
- spin_lock_bh(&efx->filter_lock);
-
- for (table_id = EFX_FARCH_FILTER_TABLE_RX_IP;
- table_id <= EFX_FARCH_FILTER_TABLE_RX_DEF;
- table_id++) {
- table = &state->table[table_id];
-
- for (filter_idx = 0; filter_idx < table->size; filter_idx++) {
- if (!test_bit(filter_idx, table->used_bitmap) ||
- table->spec[filter_idx].dmaq_id >=
- efx->n_rx_channels)
- continue;
-
- if (efx->rx_scatter)
- table->spec[filter_idx].flags |=
- EFX_FILTER_FLAG_RX_SCATTER;
- else
- table->spec[filter_idx].flags &=
- ~EFX_FILTER_FLAG_RX_SCATTER;
-
- if (table_id == EFX_FARCH_FILTER_TABLE_RX_DEF)
- /* Pushed by efx_farch_filter_push_rx_config() */
- continue;
-
- efx_farch_filter_build(&filter, &table->spec[filter_idx]);
- efx_writeo(efx, &filter,
- table->offset + table->step * filter_idx);
- }
- }
-
- efx_farch_filter_push_rx_config(efx);
-
- spin_unlock_bh(&efx->filter_lock);
-}
-
-#ifdef CONFIG_RFS_ACCEL
-
-int efx_filter_rfs(struct net_device *net_dev, const struct sk_buff *skb,
- u16 rxq_index, u32 flow_id)
-{
- struct efx_nic *efx = netdev_priv(net_dev);
- struct efx_channel *channel;
- struct efx_filter_spec spec;
- const struct iphdr *ip;
- const __be16 *ports;
- int nhoff;
- int rc;
-
- nhoff = skb_network_offset(skb);
-
- if (skb->protocol == htons(ETH_P_8021Q)) {
- EFX_BUG_ON_PARANOID(skb_headlen(skb) <
- nhoff + sizeof(struct vlan_hdr));
- if (((const struct vlan_hdr *)skb->data + nhoff)->
- h_vlan_encapsulated_proto != htons(ETH_P_IP))
- return -EPROTONOSUPPORT;
-
- /* This is IP over 802.1q VLAN. We can't filter on the
- * IP 5-tuple and the vlan together, so just strip the
- * vlan header and filter on the IP part.
- */
- nhoff += sizeof(struct vlan_hdr);
- } else if (skb->protocol != htons(ETH_P_IP)) {
- return -EPROTONOSUPPORT;
- }
-
- /* RFS must validate the IP header length before calling us */
- EFX_BUG_ON_PARANOID(skb_headlen(skb) < nhoff + sizeof(*ip));
- ip = (const struct iphdr *)(skb->data + nhoff);
- if (ip_is_fragment(ip))
- return -EPROTONOSUPPORT;
- EFX_BUG_ON_PARANOID(skb_headlen(skb) < nhoff + 4 * ip->ihl + 4);
- ports = (const __be16 *)(skb->data + nhoff + 4 * ip->ihl);
-
- efx_filter_init_rx(&spec, EFX_FILTER_PRI_HINT,
- efx->rx_scatter ? EFX_FILTER_FLAG_RX_SCATTER : 0,
- rxq_index);
- rc = efx_filter_set_ipv4_full(&spec, ip->protocol,
- ip->daddr, ports[1], ip->saddr, ports[0]);
- if (rc)
- return rc;
-
- rc = efx_filter_insert_filter(efx, &spec, true);
- if (rc < 0)
- return rc;
-
- /* Remember this so we can check whether to expire the filter later */
- efx->rps_flow_id[rc] = flow_id;
- channel = efx_get_channel(efx, skb_get_rx_queue(skb));
- ++channel->rfs_filters_added;
-
- netif_info(efx, rx_status, efx->net_dev,
- "steering %s %pI4:%u:%pI4:%u to queue %u [flow %u filter %d]\n",
- (ip->protocol == IPPROTO_TCP) ? "TCP" : "UDP",
- &ip->saddr, ntohs(ports[0]), &ip->daddr, ntohs(ports[1]),
- rxq_index, flow_id, rc);
-
- return rc;
-}
-
-bool __efx_filter_rfs_expire(struct efx_nic *efx, unsigned quota)
-{
- struct efx_farch_filter_state *state = efx->filter_state;
- struct efx_farch_filter_table *table =
- &state->table[EFX_FARCH_FILTER_TABLE_RX_IP];
- unsigned mask = table->size - 1;
- unsigned index;
- unsigned stop;
-
- if (!spin_trylock_bh(&efx->filter_lock))
- return false;
-
- index = efx->rps_expire_index;
- stop = (index + quota) & mask;
-
- while (index != stop) {
- if (test_bit(index, table->used_bitmap) &&
- table->spec[index].priority == EFX_FILTER_PRI_HINT &&
- rps_may_expire_flow(efx->net_dev,
- table->spec[index].dmaq_id,
- efx->rps_flow_id[index], index)) {
- netif_info(efx, rx_status, efx->net_dev,
- "expiring filter %d [flow %u]\n",
- index, efx->rps_flow_id[index]);
- efx_farch_filter_table_clear_entry(efx, table, index);
- }
- index = (index + 1) & mask;
- }
-
- efx->rps_expire_index = stop;
-
- spin_unlock_bh(&efx->filter_lock);
- return true;
-}
-
-#endif /* CONFIG_RFS_ACCEL */
diff --git a/drivers/net/ethernet/sfc/net_driver.h b/drivers/net/ethernet/sfc/net_driver.h
index 5287a3c..d35ce14 100644
--- a/drivers/net/ethernet/sfc/net_driver.h
+++ b/drivers/net/ethernet/sfc/net_driver.h
@@ -30,6 +30,7 @@
#include "enum.h"
#include "bitfield.h"
+#include "filter.h"
/**************************************************************************
*
@@ -1025,6 +1026,24 @@ static inline unsigned int efx_port_num(struct efx_nic *efx)
* @ev_process: Process events for a queue, up to the given NAPI quota
* @ev_read_ack: Acknowledge read events on a queue, rearming its IRQ
* @ev_test_generate: Generate a test event
+ * @filter_table_probe: Probe filter capabilities and set up filter software state
+ * @filter_table_restore: Restore filters removed from hardware
+ * @filter_table_remove: Remove filters from hardware and tear down software state
+ * @filter_update_rx_scatter: Update filters after change to rx scatter setting
+ * @filter_insert: add or replace a filter
+ * @filter_remove_safe: remove a filter by ID, carefully
+ * @filter_get_safe: retrieve a filter by ID, carefully
+ * @filter_clear_rx: remove RX filters by priority
+ * @filter_count_rx_used: Get the number of filters in use at a given priority
+ * @filter_get_rx_id_limit: Get maximum value of a filter id, plus 1
+ * @filter_get_rx_ids: Get list of RX filters at a given priority
+ * @filter_rfs_insert: Add or replace a filter for RFS. This must be
+ * atomic. The hardware change may be asynchronous but should
+ * not be delayed for long. It may fail if this can't be done
+ * atomically.
+ * @filter_rfs_expire_one: Consider expiring a filter inserted for RFS.
+ * This must check whether the specified table entry is used by RFS
+ * and that rps_may_expire_flow() returns true for it.
* @revision: Hardware architecture revision
* @txd_ptr_tbl_base: TX descriptor ring base address
* @rxd_ptr_tbl_base: RX descriptor ring base address
@@ -1102,6 +1121,32 @@ struct efx_nic_type {
int (*ev_process)(struct efx_channel *channel, int quota);
void (*ev_read_ack)(struct efx_channel *channel);
void (*ev_test_generate)(struct efx_channel *channel);
+ int (*filter_table_probe)(struct efx_nic *efx);
+ void (*filter_table_restore)(struct efx_nic *efx);
+ void (*filter_table_remove)(struct efx_nic *efx);
+ void (*filter_update_rx_scatter)(struct efx_nic *efx);
+ s32 (*filter_insert)(struct efx_nic *efx,
+ struct efx_filter_spec *spec, bool replace);
+ int (*filter_remove_safe)(struct efx_nic *efx,
+ enum efx_filter_priority priority,
+ u32 filter_id);
+ int (*filter_get_safe)(struct efx_nic *efx,
+ enum efx_filter_priority priority,
+ u32 filter_id, struct efx_filter_spec *);
+ void (*filter_clear_rx)(struct efx_nic *efx,
+ enum efx_filter_priority priority);
+ u32 (*filter_count_rx_used)(struct efx_nic *efx,
+ enum efx_filter_priority priority);
+ u32 (*filter_get_rx_id_limit)(struct efx_nic *efx);
+ s32 (*filter_get_rx_ids)(struct efx_nic *efx,
+ enum efx_filter_priority priority,
+ u32 *buf, u32 size);
+#ifdef CONFIG_RFS_ACCEL
+ s32 (*filter_rfs_insert)(struct efx_nic *efx,
+ struct efx_filter_spec *spec);
+ bool (*filter_rfs_expire_one)(struct efx_nic *efx, u32 flow_id,
+ unsigned int index);
+#endif
int revision;
unsigned int txd_ptr_tbl_base;
@@ -1117,6 +1162,7 @@ struct efx_nic_type {
unsigned int timer_period_max;
netdev_features_t offload_features;
int mcdi_max_ver;
+ unsigned int max_rx_ip_filters;
};
/**************************************************************************
diff --git a/drivers/net/ethernet/sfc/nic.h b/drivers/net/ethernet/sfc/nic.h
index 25e25b6..69298c9 100644
--- a/drivers/net/ethernet/sfc/nic.h
+++ b/drivers/net/ethernet/sfc/nic.h
@@ -404,6 +404,34 @@ extern int efx_farch_ev_process(struct efx_channel *channel, int quota);
extern void efx_farch_ev_read_ack(struct efx_channel *channel);
extern void efx_farch_ev_test_generate(struct efx_channel *channel);
+/* Falcon/Siena filter operations */
+extern int efx_farch_filter_table_probe(struct efx_nic *efx);
+extern void efx_farch_filter_table_restore(struct efx_nic *efx);
+extern void efx_farch_filter_table_remove(struct efx_nic *efx);
+extern void efx_farch_filter_update_rx_scatter(struct efx_nic *efx);
+extern s32 efx_farch_filter_insert(struct efx_nic *efx,
+ struct efx_filter_spec *spec, bool replace);
+extern int efx_farch_filter_remove_safe(struct efx_nic *efx,
+ enum efx_filter_priority priority,
+ u32 filter_id);
+extern int efx_farch_filter_get_safe(struct efx_nic *efx,
+ enum efx_filter_priority priority,
+ u32 filter_id, struct efx_filter_spec *);
+extern void efx_farch_filter_clear_rx(struct efx_nic *efx,
+ enum efx_filter_priority priority);
+extern u32 efx_farch_filter_count_rx_used(struct efx_nic *efx,
+ enum efx_filter_priority priority);
+extern u32 efx_farch_filter_get_rx_id_limit(struct efx_nic *efx);
+extern s32 efx_farch_filter_get_rx_ids(struct efx_nic *efx,
+ enum efx_filter_priority priority,
+ u32 *buf, u32 size);
+#ifdef CONFIG_RFS_ACCEL
+extern s32 efx_farch_filter_rfs_insert(struct efx_nic *efx,
+ struct efx_filter_spec *spec);
+extern bool efx_farch_filter_rfs_expire_one(struct efx_nic *efx, u32 flow_id,
+ unsigned int index);
+#endif
+
extern bool efx_nic_event_present(struct efx_channel *channel);
/* Some statistics are computed as A - B where A and B each increase
diff --git a/drivers/net/ethernet/sfc/rx.c b/drivers/net/ethernet/sfc/rx.c
index f2b78cd..1299092 100644
--- a/drivers/net/ethernet/sfc/rx.c
+++ b/drivers/net/ethernet/sfc/rx.c
@@ -21,6 +21,7 @@
#include <net/checksum.h>
#include "net_driver.h"
#include "efx.h"
+#include "filter.h"
#include "nic.h"
#include "selftest.h"
#include "workarounds.h"
@@ -802,3 +803,96 @@ module_param(rx_refill_threshold, uint, 0444);
MODULE_PARM_DESC(rx_refill_threshold,
"RX descriptor ring refill threshold (%)");
+#ifdef CONFIG_RFS_ACCEL
+
+int efx_filter_rfs(struct net_device *net_dev, const struct sk_buff *skb,
+ u16 rxq_index, u32 flow_id)
+{
+ struct efx_nic *efx = netdev_priv(net_dev);
+ struct efx_channel *channel;
+ struct efx_filter_spec spec;
+ const struct iphdr *ip;
+ const __be16 *ports;
+ int nhoff;
+ int rc;
+
+ nhoff = skb_network_offset(skb);
+
+ if (skb->protocol == htons(ETH_P_8021Q)) {
+ EFX_BUG_ON_PARANOID(skb_headlen(skb) <
+ nhoff + sizeof(struct vlan_hdr));
+ if (((const struct vlan_hdr *)skb->data + nhoff)->
+ h_vlan_encapsulated_proto != htons(ETH_P_IP))
+ return -EPROTONOSUPPORT;
+
+ /* This is IP over 802.1q VLAN. We can't filter on the
+ * IP 5-tuple and the vlan together, so just strip the
+ * vlan header and filter on the IP part.
+ */
+ nhoff += sizeof(struct vlan_hdr);
+ } else if (skb->protocol != htons(ETH_P_IP)) {
+ return -EPROTONOSUPPORT;
+ }
+
+ /* RFS must validate the IP header length before calling us */
+ EFX_BUG_ON_PARANOID(skb_headlen(skb) < nhoff + sizeof(*ip));
+ ip = (const struct iphdr *)(skb->data + nhoff);
+ if (ip_is_fragment(ip))
+ return -EPROTONOSUPPORT;
+ EFX_BUG_ON_PARANOID(skb_headlen(skb) < nhoff + 4 * ip->ihl + 4);
+ ports = (const __be16 *)(skb->data + nhoff + 4 * ip->ihl);
+
+ efx_filter_init_rx(&spec, EFX_FILTER_PRI_HINT,
+ efx->rx_scatter ? EFX_FILTER_FLAG_RX_SCATTER : 0,
+ rxq_index);
+ rc = efx_filter_set_ipv4_full(&spec, ip->protocol,
+ ip->daddr, ports[1], ip->saddr, ports[0]);
+ if (rc)
+ return rc;
+
+ rc = efx->type->filter_rfs_insert(efx, &spec);
+ if (rc < 0)
+ return rc;
+
+ /* Remember this so we can check whether to expire the filter later */
+ efx->rps_flow_id[rc] = flow_id;
+ channel = efx_get_channel(efx, skb_get_rx_queue(skb));
+ ++channel->rfs_filters_added;
+
+ netif_info(efx, rx_status, efx->net_dev,
+ "steering %s %pI4:%u:%pI4:%u to queue %u [flow %u filter %d]\n",
+ (ip->protocol == IPPROTO_TCP) ? "TCP" : "UDP",
+ &ip->saddr, ntohs(ports[0]), &ip->daddr, ntohs(ports[1]),
+ rxq_index, flow_id, rc);
+
+ return rc;
+}
+
+bool __efx_filter_rfs_expire(struct efx_nic *efx, unsigned int quota)
+{
+ bool (*expire_one)(struct efx_nic *efx, u32 flow_id, unsigned int index);
+ unsigned int index, size;
+ u32 flow_id;
+
+ if (!spin_trylock_bh(&efx->filter_lock))
+ return false;
+
+ expire_one = efx->type->filter_rfs_expire_one;
+ index = efx->rps_expire_index;
+ size = efx->type->max_rx_ip_filters;
+ while (quota--) {
+ flow_id = efx->rps_flow_id[index];
+ if (expire_one(efx, flow_id, index))
+ netif_info(efx, rx_status, efx->net_dev,
+ "expired filter %d [flow %u]\n",
+ index, flow_id);
+ if (++index == size)
+ index = 0;
+ }
+ efx->rps_expire_index = index;
+
+ spin_unlock_bh(&efx->filter_lock);
+ return true;
+}
+
+#endif /* CONFIG_RFS_ACCEL */
diff --git a/drivers/net/ethernet/sfc/siena.c b/drivers/net/ethernet/sfc/siena.c
index 23e5731..5120cd8 100644
--- a/drivers/net/ethernet/sfc/siena.c
+++ b/drivers/net/ethernet/sfc/siena.c
@@ -736,6 +736,21 @@ const struct efx_nic_type siena_a0_nic_type = {
.ev_process = efx_farch_ev_process,
.ev_read_ack = efx_farch_ev_read_ack,
.ev_test_generate = efx_farch_ev_test_generate,
+ .filter_table_probe = efx_farch_filter_table_probe,
+ .filter_table_restore = efx_farch_filter_table_restore,
+ .filter_table_remove = efx_farch_filter_table_remove,
+ .filter_update_rx_scatter = efx_farch_filter_update_rx_scatter,
+ .filter_insert = efx_farch_filter_insert,
+ .filter_remove_safe = efx_farch_filter_remove_safe,
+ .filter_get_safe = efx_farch_filter_get_safe,
+ .filter_clear_rx = efx_farch_filter_clear_rx,
+ .filter_count_rx_used = efx_farch_filter_count_rx_used,
+ .filter_get_rx_id_limit = efx_farch_filter_get_rx_id_limit,
+ .filter_get_rx_ids = efx_farch_filter_get_rx_ids,
+#ifdef CONFIG_RFS_ACCEL
+ .filter_rfs_insert = efx_farch_filter_rfs_insert,
+ .filter_rfs_expire_one = efx_farch_filter_rfs_expire_one,
+#endif
.revision = EFX_REV_SIENA_A0,
.txd_ptr_tbl_base = FR_BZ_TX_DESC_PTR_TBL,
@@ -752,4 +767,5 @@ const struct efx_nic_type siena_a0_nic_type = {
.offload_features = (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
NETIF_F_RXHASH | NETIF_F_NTUPLE),
.mcdi_max_ver = 1,
+ .max_rx_ip_filters = FR_BZ_RX_FILTER_TBL0_ROWS,
};
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [PATCH net-next 09/16] sfc: Refactor Falcon-arch search limit reset
From: Ben Hutchings @ 2013-08-27 20:46 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377635844.13272.69.camel@bwh-desktop.uk.level5networks.com>
Currently every call to efx_farch_filter_table_clear_entry() is
shortly followed by a conditional reset of the table limits. The new
limits (0) are not pushed to hardware until the next filter insertion.
Move both the reset and the hardware reconfiguration into
efx_farch_filter_table_clear_entry(), and add an explanatory comment.
Also, make consistent use of the term 'search limit' for the maximum
number of probes the NIC must make when searching for a filter of a
particular type.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/filter.c | 59 ++++++++++++++++++++-------------------
1 file changed, 30 insertions(+), 29 deletions(-)
diff --git a/drivers/net/ethernet/sfc/filter.c b/drivers/net/ethernet/sfc/filter.c
index 29986d3..ad66376 100644
--- a/drivers/net/ethernet/sfc/filter.c
+++ b/drivers/net/ethernet/sfc/filter.c
@@ -22,7 +22,7 @@
#define EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD 3
#define EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL 1
-/* Hard maximum hop limit. Hardware will time-out beyond 200-something.
+/* Hard maximum search limit. Hardware will time-out beyond 200-something.
* We also need to avoid infinite loops in efx_farch_filter_search() when the
* table is full.
*/
@@ -74,7 +74,7 @@ struct efx_farch_filter_table {
unsigned used; /* number currently used */
unsigned long *used_bitmap;
struct efx_farch_filter_spec *spec;
- unsigned search_depth[EFX_FARCH_FILTER_TYPE_COUNT];
+ unsigned search_limit[EFX_FARCH_FILTER_TYPE_COUNT];
};
struct efx_farch_filter_state {
@@ -129,12 +129,6 @@ efx_farch_filter_spec_table_id(const struct efx_farch_filter_spec *spec)
return (spec->type >> 2) + ((spec->flags & EFX_FILTER_FLAG_TX) ? 2 : 0);
}
-static void
-efx_farch_filter_table_reset_search_depth(struct efx_farch_filter_table *table)
-{
- memset(table->search_depth, 0, sizeof(table->search_depth));
-}
-
static void efx_farch_filter_push_rx_config(struct efx_nic *efx)
{
struct efx_farch_filter_state *state = efx->filter_state;
@@ -145,27 +139,27 @@ static void efx_farch_filter_push_rx_config(struct efx_nic *efx)
table = &state->table[EFX_FARCH_FILTER_TABLE_RX_IP];
EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_TCP_FULL_SRCH_LIMIT,
- table->search_depth[EFX_FARCH_FILTER_TCP_FULL] +
+ table->search_limit[EFX_FARCH_FILTER_TCP_FULL] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_TCP_WILD_SRCH_LIMIT,
- table->search_depth[EFX_FARCH_FILTER_TCP_WILD] +
+ table->search_limit[EFX_FARCH_FILTER_TCP_WILD] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_UDP_FULL_SRCH_LIMIT,
- table->search_depth[EFX_FARCH_FILTER_UDP_FULL] +
+ table->search_limit[EFX_FARCH_FILTER_UDP_FULL] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_UDP_WILD_SRCH_LIMIT,
- table->search_depth[EFX_FARCH_FILTER_UDP_WILD] +
+ table->search_limit[EFX_FARCH_FILTER_UDP_WILD] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
table = &state->table[EFX_FARCH_FILTER_TABLE_RX_MAC];
if (table->size) {
EFX_SET_OWORD_FIELD(
filter_ctl, FRF_CZ_ETHERNET_FULL_SEARCH_LIMIT,
- table->search_depth[EFX_FARCH_FILTER_MAC_FULL] +
+ table->search_limit[EFX_FARCH_FILTER_MAC_FULL] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
EFX_SET_OWORD_FIELD(
filter_ctl, FRF_CZ_ETHERNET_WILDCARD_SEARCH_LIMIT,
- table->search_depth[EFX_FARCH_FILTER_MAC_WILD] +
+ table->search_limit[EFX_FARCH_FILTER_MAC_WILD] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
}
@@ -221,11 +215,11 @@ static void efx_farch_filter_push_tx_limits(struct efx_nic *efx)
if (table->size) {
EFX_SET_OWORD_FIELD(
tx_cfg, FRF_CZ_TX_ETH_FILTER_FULL_SEARCH_RANGE,
- table->search_depth[EFX_FARCH_FILTER_MAC_FULL] +
+ table->search_limit[EFX_FARCH_FILTER_MAC_FULL] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
EFX_SET_OWORD_FIELD(
tx_cfg, FRF_CZ_TX_ETH_FILTER_WILD_SEARCH_RANGE,
- table->search_depth[EFX_FARCH_FILTER_MAC_WILD] +
+ table->search_limit[EFX_FARCH_FILTER_MAC_WILD] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
}
@@ -633,8 +627,8 @@ s32 efx_filter_insert_filter(struct efx_nic *efx,
return -EINVAL;
netif_vdbg(efx, hw, efx->net_dev,
- "%s: type %d search_depth=%d", __func__, spec.type,
- table->search_depth[spec.type]);
+ "%s: type %d search_limit=%d", __func__, spec.type,
+ table->search_limit[spec.type]);
if (table->id == EFX_FARCH_FILTER_TABLE_RX_DEF) {
/* One filter spec per type */
@@ -664,7 +658,7 @@ s32 efx_filter_insert_filter(struct efx_nic *efx,
u32 key = efx_farch_filter_build(&filter, &spec);
unsigned int hash = efx_farch_filter_hash(key);
unsigned int incr = efx_farch_filter_increment(key);
- unsigned int max_rep_depth = table->search_depth[spec.type];
+ unsigned int max_rep_depth = table->search_limit[spec.type];
unsigned int max_ins_depth =
spec.priority <= EFX_FILTER_PRI_HINT ?
EFX_FARCH_FILTER_CTL_SRCH_HINT_MAX :
@@ -732,8 +726,8 @@ s32 efx_filter_insert_filter(struct efx_nic *efx,
if (table->id == EFX_FARCH_FILTER_TABLE_RX_DEF) {
efx_farch_filter_push_rx_config(efx);
} else {
- if (table->search_depth[spec.type] < depth) {
- table->search_depth[spec.type] = depth;
+ if (table->search_limit[spec.type] < depth) {
+ table->search_limit[spec.type] = depth;
if (spec.flags & EFX_FILTER_FLAG_TX)
efx_farch_filter_push_tx_limits(efx);
else
@@ -779,6 +773,21 @@ efx_farch_filter_table_clear_entry(struct efx_nic *efx,
efx_writeo(efx, &filter,
table->offset + table->step * filter_idx);
+
+ /* If this filter required a greater search depth than
+ * any other, the search limit for its type can now be
+ * decreased. However, it is hard to determine that
+ * unless the table has become completely empty - in
+ * which case, all its search limits can be set to 0.
+ */
+ if (unlikely(table->used == 0)) {
+ memset(table->search_limit, 0,
+ sizeof(table->search_limit));
+ if (table->id == EFX_FARCH_FILTER_TABLE_TX_MAC)
+ efx_farch_filter_push_tx_limits(efx);
+ else
+ efx_farch_filter_push_rx_config(efx);
+ }
}
}
@@ -817,8 +826,6 @@ int efx_filter_remove_id_safe(struct efx_nic *efx,
if (test_bit(filter_idx, table->used_bitmap) &&
spec->priority == priority) {
efx_farch_filter_table_clear_entry(efx, table, filter_idx);
- if (table->used == 0)
- efx_farch_filter_table_reset_search_depth(table);
rc = 0;
} else {
rc = -ENOENT;
@@ -885,14 +892,10 @@ efx_farch_filter_table_clear(struct efx_nic *efx,
unsigned int filter_idx;
spin_lock_bh(&efx->filter_lock);
-
for (filter_idx = 0; filter_idx < table->size; ++filter_idx)
if (table->spec[filter_idx].priority <= priority)
efx_farch_filter_table_clear_entry(efx, table,
filter_idx);
- if (table->used == 0)
- efx_farch_filter_table_reset_search_depth(table);
-
spin_unlock_bh(&efx->filter_lock);
}
@@ -1233,8 +1236,6 @@ bool __efx_filter_rfs_expire(struct efx_nic *efx, unsigned quota)
}
efx->rps_expire_index = stop;
- if (table->used == 0)
- efx_farch_filter_table_reset_search_depth(table);
spin_unlock_bh(&efx->filter_lock);
return true;
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [PATCH net-next 08/16] sfc: Split Falcon-arch-specific and common filter state
From: Ben Hutchings @ 2013-08-27 20:45 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377635844.13272.69.camel@bwh-desktop.uk.level5networks.com>
Move the common state from struct efx_filter_state into struct efx_nic.
Rename struct efx_filter_state to efx_farch_filter_state and change
the type of efx_nic::filter_state to void *.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/filter.c | 100 ++++++++++++++++------------------
drivers/net/ethernet/sfc/net_driver.h | 13 ++++-
2 files changed, 58 insertions(+), 55 deletions(-)
diff --git a/drivers/net/ethernet/sfc/filter.c b/drivers/net/ethernet/sfc/filter.c
index 96f8f2e..29986d3 100644
--- a/drivers/net/ethernet/sfc/filter.c
+++ b/drivers/net/ethernet/sfc/filter.c
@@ -77,13 +77,8 @@ struct efx_farch_filter_table {
unsigned search_depth[EFX_FARCH_FILTER_TYPE_COUNT];
};
-struct efx_filter_state {
- spinlock_t lock;
+struct efx_farch_filter_state {
struct efx_farch_filter_table table[EFX_FARCH_FILTER_TABLE_COUNT];
-#ifdef CONFIG_RFS_ACCEL
- u32 *rps_flow_id;
- unsigned rps_expire_index;
-#endif
};
static void
@@ -142,7 +137,7 @@ efx_farch_filter_table_reset_search_depth(struct efx_farch_filter_table *table)
static void efx_farch_filter_push_rx_config(struct efx_nic *efx)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
struct efx_farch_filter_table *table;
efx_oword_t filter_ctl;
@@ -216,7 +211,7 @@ static void efx_farch_filter_push_rx_config(struct efx_nic *efx)
static void efx_farch_filter_push_tx_limits(struct efx_nic *efx)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
struct efx_farch_filter_table *table;
efx_oword_t tx_cfg;
@@ -427,7 +422,7 @@ efx_farch_filter_to_gen_spec(struct efx_filter_spec *gen_spec,
static void
efx_farch_filter_reset_rx_def(struct efx_nic *efx, unsigned filter_idx)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
struct efx_farch_filter_table *table =
&state->table[EFX_FARCH_FILTER_TABLE_RX_DEF];
struct efx_farch_filter_spec *spec = &table->spec[filter_idx];
@@ -585,7 +580,7 @@ static inline unsigned int efx_farch_filter_id_index(u32 id)
u32 efx_filter_get_rx_id_limit(struct efx_nic *efx)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
unsigned int range = EFX_FARCH_FILTER_MATCH_PRI_COUNT - 1;
enum efx_farch_filter_table_id table_id;
@@ -621,7 +616,7 @@ s32 efx_filter_insert_filter(struct efx_nic *efx,
struct efx_filter_spec *gen_spec,
bool replace_equal)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
struct efx_farch_filter_table *table;
struct efx_farch_filter_spec spec;
efx_oword_t filter;
@@ -649,7 +644,7 @@ s32 efx_filter_insert_filter(struct efx_nic *efx,
rep_index = spec.type - EFX_FARCH_FILTER_UC_DEF;
ins_index = rep_index;
- spin_lock_bh(&state->lock);
+ spin_lock_bh(&efx->filter_lock);
} else {
/* Search concurrently for
* (1) a filter to be replaced (rep_index): any filter
@@ -679,7 +674,7 @@ s32 efx_filter_insert_filter(struct efx_nic *efx,
ins_index = -1;
depth = 1;
- spin_lock_bh(&state->lock);
+ spin_lock_bh(&efx->filter_lock);
for (;;) {
if (!test_bit(i, table->used_bitmap)) {
@@ -762,7 +757,7 @@ s32 efx_filter_insert_filter(struct efx_nic *efx,
rc = efx_farch_filter_make_id(&spec, ins_index);
out:
- spin_unlock_bh(&state->lock);
+ spin_unlock_bh(&efx->filter_lock);
return rc;
}
@@ -800,7 +795,7 @@ int efx_filter_remove_id_safe(struct efx_nic *efx,
enum efx_filter_priority priority,
u32 filter_id)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
enum efx_farch_filter_table_id table_id;
struct efx_farch_filter_table *table;
unsigned int filter_idx;
@@ -817,7 +812,7 @@ int efx_filter_remove_id_safe(struct efx_nic *efx,
return -ENOENT;
spec = &table->spec[filter_idx];
- spin_lock_bh(&state->lock);
+ spin_lock_bh(&efx->filter_lock);
if (test_bit(filter_idx, table->used_bitmap) &&
spec->priority == priority) {
@@ -829,7 +824,7 @@ int efx_filter_remove_id_safe(struct efx_nic *efx,
rc = -ENOENT;
}
- spin_unlock_bh(&state->lock);
+ spin_unlock_bh(&efx->filter_lock);
return rc;
}
@@ -848,7 +843,7 @@ int efx_filter_get_filter_safe(struct efx_nic *efx,
enum efx_filter_priority priority,
u32 filter_id, struct efx_filter_spec *spec_buf)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
enum efx_farch_filter_table_id table_id;
struct efx_farch_filter_table *table;
struct efx_farch_filter_spec *spec;
@@ -865,7 +860,7 @@ int efx_filter_get_filter_safe(struct efx_nic *efx,
return -ENOENT;
spec = &table->spec[filter_idx];
- spin_lock_bh(&state->lock);
+ spin_lock_bh(&efx->filter_lock);
if (test_bit(filter_idx, table->used_bitmap) &&
spec->priority == priority) {
@@ -875,7 +870,7 @@ int efx_filter_get_filter_safe(struct efx_nic *efx,
rc = -ENOENT;
}
- spin_unlock_bh(&state->lock);
+ spin_unlock_bh(&efx->filter_lock);
return rc;
}
@@ -885,11 +880,11 @@ efx_farch_filter_table_clear(struct efx_nic *efx,
enum efx_farch_filter_table_id table_id,
enum efx_filter_priority priority)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
struct efx_farch_filter_table *table = &state->table[table_id];
unsigned int filter_idx;
- spin_lock_bh(&state->lock);
+ spin_lock_bh(&efx->filter_lock);
for (filter_idx = 0; filter_idx < table->size; ++filter_idx)
if (table->spec[filter_idx].priority <= priority)
@@ -898,7 +893,7 @@ efx_farch_filter_table_clear(struct efx_nic *efx,
if (table->used == 0)
efx_farch_filter_table_reset_search_depth(table);
- spin_unlock_bh(&state->lock);
+ spin_unlock_bh(&efx->filter_lock);
}
/**
@@ -917,13 +912,13 @@ void efx_filter_clear_rx(struct efx_nic *efx, enum efx_filter_priority priority)
u32 efx_filter_count_rx_used(struct efx_nic *efx,
enum efx_filter_priority priority)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
enum efx_farch_filter_table_id table_id;
struct efx_farch_filter_table *table;
unsigned int filter_idx;
u32 count = 0;
- spin_lock_bh(&state->lock);
+ spin_lock_bh(&efx->filter_lock);
for (table_id = EFX_FARCH_FILTER_TABLE_RX_IP;
table_id <= EFX_FARCH_FILTER_TABLE_RX_DEF;
@@ -936,7 +931,7 @@ u32 efx_filter_count_rx_used(struct efx_nic *efx,
}
}
- spin_unlock_bh(&state->lock);
+ spin_unlock_bh(&efx->filter_lock);
return count;
}
@@ -945,13 +940,13 @@ s32 efx_filter_get_rx_ids(struct efx_nic *efx,
enum efx_filter_priority priority,
u32 *buf, u32 size)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
enum efx_farch_filter_table_id table_id;
struct efx_farch_filter_table *table;
unsigned int filter_idx;
s32 count = 0;
- spin_lock_bh(&state->lock);
+ spin_lock_bh(&efx->filter_lock);
for (table_id = EFX_FARCH_FILTER_TABLE_RX_IP;
table_id <= EFX_FARCH_FILTER_TABLE_RX_DEF;
@@ -970,7 +965,7 @@ s32 efx_filter_get_rx_ids(struct efx_nic *efx,
}
}
out:
- spin_unlock_bh(&state->lock);
+ spin_unlock_bh(&efx->filter_lock);
return count;
}
@@ -978,13 +973,13 @@ out:
/* Restore filter stater after reset */
void efx_restore_filters(struct efx_nic *efx)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
enum efx_farch_filter_table_id table_id;
struct efx_farch_filter_table *table;
efx_oword_t filter;
unsigned int filter_idx;
- spin_lock_bh(&state->lock);
+ spin_lock_bh(&efx->filter_lock);
for (table_id = 0; table_id < EFX_FARCH_FILTER_TABLE_COUNT; table_id++) {
table = &state->table[table_id];
@@ -1005,28 +1000,28 @@ void efx_restore_filters(struct efx_nic *efx)
efx_farch_filter_push_rx_config(efx);
efx_farch_filter_push_tx_limits(efx);
- spin_unlock_bh(&state->lock);
+ spin_unlock_bh(&efx->filter_lock);
}
int efx_probe_filters(struct efx_nic *efx)
{
- struct efx_filter_state *state;
+ struct efx_farch_filter_state *state;
struct efx_farch_filter_table *table;
unsigned table_id;
- state = kzalloc(sizeof(*efx->filter_state), GFP_KERNEL);
+ state = kzalloc(sizeof(struct efx_farch_filter_state), GFP_KERNEL);
if (!state)
return -ENOMEM;
efx->filter_state = state;
- spin_lock_init(&state->lock);
+ spin_lock_init(&efx->filter_lock);
if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0) {
#ifdef CONFIG_RFS_ACCEL
- state->rps_flow_id = kcalloc(FR_BZ_RX_FILTER_TBL0_ROWS,
- sizeof(*state->rps_flow_id),
- GFP_KERNEL);
- if (!state->rps_flow_id)
+ efx->rps_flow_id = kcalloc(FR_BZ_RX_FILTER_TBL0_ROWS,
+ sizeof(*efx->rps_flow_id),
+ GFP_KERNEL);
+ if (!efx->rps_flow_id)
goto fail;
#endif
table = &state->table[EFX_FARCH_FILTER_TABLE_RX_IP];
@@ -1086,7 +1081,7 @@ fail:
void efx_remove_filters(struct efx_nic *efx)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
enum efx_farch_filter_table_id table_id;
for (table_id = 0; table_id < EFX_FARCH_FILTER_TABLE_COUNT; table_id++) {
@@ -1094,7 +1089,7 @@ void efx_remove_filters(struct efx_nic *efx)
vfree(state->table[table_id].spec);
}
#ifdef CONFIG_RFS_ACCEL
- kfree(state->rps_flow_id);
+ kfree(efx->rps_flow_id);
#endif
kfree(state);
}
@@ -1102,13 +1097,13 @@ void efx_remove_filters(struct efx_nic *efx)
/* Update scatter enable flags for filters pointing to our own RX queues */
void efx_filter_update_rx_scatter(struct efx_nic *efx)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
enum efx_farch_filter_table_id table_id;
struct efx_farch_filter_table *table;
efx_oword_t filter;
unsigned int filter_idx;
- spin_lock_bh(&state->lock);
+ spin_lock_bh(&efx->filter_lock);
for (table_id = EFX_FARCH_FILTER_TABLE_RX_IP;
table_id <= EFX_FARCH_FILTER_TABLE_RX_DEF;
@@ -1140,7 +1135,7 @@ void efx_filter_update_rx_scatter(struct efx_nic *efx)
efx_farch_filter_push_rx_config(efx);
- spin_unlock_bh(&state->lock);
+ spin_unlock_bh(&efx->filter_lock);
}
#ifdef CONFIG_RFS_ACCEL
@@ -1150,7 +1145,6 @@ int efx_filter_rfs(struct net_device *net_dev, const struct sk_buff *skb,
{
struct efx_nic *efx = netdev_priv(net_dev);
struct efx_channel *channel;
- struct efx_filter_state *state = efx->filter_state;
struct efx_filter_spec spec;
const struct iphdr *ip;
const __be16 *ports;
@@ -1196,7 +1190,7 @@ int efx_filter_rfs(struct net_device *net_dev, const struct sk_buff *skb,
return rc;
/* Remember this so we can check whether to expire the filter later */
- state->rps_flow_id[rc] = flow_id;
+ efx->rps_flow_id[rc] = flow_id;
channel = efx_get_channel(efx, skb_get_rx_queue(skb));
++channel->rfs_filters_added;
@@ -1211,17 +1205,17 @@ int efx_filter_rfs(struct net_device *net_dev, const struct sk_buff *skb,
bool __efx_filter_rfs_expire(struct efx_nic *efx, unsigned quota)
{
- struct efx_filter_state *state = efx->filter_state;
+ struct efx_farch_filter_state *state = efx->filter_state;
struct efx_farch_filter_table *table =
&state->table[EFX_FARCH_FILTER_TABLE_RX_IP];
unsigned mask = table->size - 1;
unsigned index;
unsigned stop;
- if (!spin_trylock_bh(&state->lock))
+ if (!spin_trylock_bh(&efx->filter_lock))
return false;
- index = state->rps_expire_index;
+ index = efx->rps_expire_index;
stop = (index + quota) & mask;
while (index != stop) {
@@ -1229,20 +1223,20 @@ bool __efx_filter_rfs_expire(struct efx_nic *efx, unsigned quota)
table->spec[index].priority == EFX_FILTER_PRI_HINT &&
rps_may_expire_flow(efx->net_dev,
table->spec[index].dmaq_id,
- state->rps_flow_id[index], index)) {
+ efx->rps_flow_id[index], index)) {
netif_info(efx, rx_status, efx->net_dev,
"expiring filter %d [flow %u]\n",
- index, state->rps_flow_id[index]);
+ index, efx->rps_flow_id[index]);
efx_farch_filter_table_clear_entry(efx, table, index);
}
index = (index + 1) & mask;
}
- state->rps_expire_index = stop;
+ efx->rps_expire_index = stop;
if (table->used == 0)
efx_farch_filter_table_reset_search_depth(table);
- spin_unlock_bh(&state->lock);
+ spin_unlock_bh(&efx->filter_lock);
return true;
}
diff --git a/drivers/net/ethernet/sfc/net_driver.h b/drivers/net/ethernet/sfc/net_driver.h
index 694c572..5287a3c 100644
--- a/drivers/net/ethernet/sfc/net_driver.h
+++ b/drivers/net/ethernet/sfc/net_driver.h
@@ -676,7 +676,6 @@ union efx_multicast_hash {
efx_oword_t oword[EFX_MCAST_HASH_ENTRIES / sizeof(efx_oword_t) / 8];
};
-struct efx_filter_state;
struct efx_vf;
struct vfdi_status;
@@ -763,6 +762,11 @@ struct vfdi_status;
* @loopback_mode: Loopback status
* @loopback_modes: Supported loopback mode bitmask
* @loopback_selftest: Offline self-test private state
+ * @filter_lock: Filter table lock
+ * @filter_state: Architecture-dependent filter table state
+ * @rps_flow_id: Flow IDs of filters allocated for accelerated RFS,
+ * indexed by filter ID
+ * @rps_expire_index: Next index to check for expiry in @rps_flow_id
* @drain_pending: Count of RX and TX queues that haven't been flushed and drained.
* @rxq_flush_pending: Count of number of receive queues that need to be flushed.
* Decremented when the efx_flush_rx_queue() is called.
@@ -898,7 +902,12 @@ struct efx_nic {
void *loopback_selftest;
- struct efx_filter_state *filter_state;
+ spinlock_t filter_lock;
+ void *filter_state;
+#ifdef CONFIG_RFS_ACCEL
+ u32 *rps_flow_id;
+ unsigned int rps_expire_index;
+#endif
atomic_t drain_pending;
atomic_t rxq_flush_pending;
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [PATCH net-next 07/16] sfc: Extend and abstract efx_filter_spec to cover Huntington/EF10
From: Ben Hutchings @ 2013-08-27 20:45 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377635844.13272.69.camel@bwh-desktop.uk.level5networks.com>
Replace type field with match_flags. Add rss_context and match values
covering of most of what is now in the MCDI protocol.
Change some fields into bitfields so that the structure size doesn't grow
beyond 64 bytes.
Ditch the filter decoding functions as it is now easier to pick apart
the abstract structure.
Rewrite ethtool NFC rule functions to set/get filter match flags and
values directly.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/ethtool.c | 211 +++++++++--------
drivers/net/ethernet/sfc/filter.c | 455 +++++++++++++++++--------------------
drivers/net/ethernet/sfc/filter.h | 220 ++++++++++++++----
3 files changed, 494 insertions(+), 392 deletions(-)
diff --git a/drivers/net/ethernet/sfc/ethtool.c b/drivers/net/ethernet/sfc/ethtool.c
index ec5cacd..58ae28b 100644
--- a/drivers/net/ethernet/sfc/ethtool.c
+++ b/drivers/net/ethernet/sfc/ethtool.c
@@ -799,11 +799,12 @@ static int efx_ethtool_reset(struct net_device *net_dev, u32 *flags)
return efx_reset(efx, rc);
}
-/* MAC address mask including only MC flag */
-static const u8 mac_addr_mc_mask[ETH_ALEN] = { 0x01, 0, 0, 0, 0, 0 };
+/* MAC address mask including only I/G bit */
+static const u8 mac_addr_ig_mask[ETH_ALEN] = { 0x01, 0, 0, 0, 0, 0 };
#define IP4_ADDR_FULL_MASK ((__force __be32)~0)
#define PORT_FULL_MASK ((__force __be16)~0)
+#define ETHER_TYPE_FULL_MASK ((__force __be16)~0)
static int efx_ethtool_get_class_rule(struct efx_nic *efx,
struct ethtool_rx_flow_spec *rule)
@@ -813,8 +814,6 @@ static int efx_ethtool_get_class_rule(struct efx_nic *efx,
struct ethhdr *mac_entry = &rule->h_u.ether_spec;
struct ethhdr *mac_mask = &rule->m_u.ether_spec;
struct efx_filter_spec spec;
- u16 vid;
- u8 proto;
int rc;
rc = efx_filter_get_filter_safe(efx, EFX_FILTER_PRI_MANUAL,
@@ -827,39 +826,67 @@ static int efx_ethtool_get_class_rule(struct efx_nic *efx,
else
rule->ring_cookie = spec.dmaq_id;
- if (spec.type == EFX_FILTER_MC_DEF || spec.type == EFX_FILTER_UC_DEF) {
- rule->flow_type = ETHER_FLOW;
- memcpy(mac_mask->h_dest, mac_addr_mc_mask, ETH_ALEN);
- if (spec.type == EFX_FILTER_MC_DEF)
- memcpy(mac_entry->h_dest, mac_addr_mc_mask, ETH_ALEN);
- return 0;
- }
-
- rc = efx_filter_get_eth_local(&spec, &vid, mac_entry->h_dest);
- if (rc == 0) {
+ if ((spec.match_flags & EFX_FILTER_MATCH_ETHER_TYPE) &&
+ spec.ether_type == htons(ETH_P_IP) &&
+ (spec.match_flags & EFX_FILTER_MATCH_IP_PROTO) &&
+ (spec.ip_proto == IPPROTO_TCP || spec.ip_proto == IPPROTO_UDP) &&
+ !(spec.match_flags &
+ ~(EFX_FILTER_MATCH_ETHER_TYPE | EFX_FILTER_MATCH_OUTER_VID |
+ EFX_FILTER_MATCH_LOC_HOST | EFX_FILTER_MATCH_REM_HOST |
+ EFX_FILTER_MATCH_IP_PROTO |
+ EFX_FILTER_MATCH_LOC_PORT | EFX_FILTER_MATCH_REM_PORT))) {
+ rule->flow_type = ((spec.ip_proto == IPPROTO_TCP) ?
+ TCP_V4_FLOW : UDP_V4_FLOW);
+ if (spec.match_flags & EFX_FILTER_MATCH_LOC_HOST) {
+ ip_entry->ip4dst = spec.loc_host[0];
+ ip_mask->ip4dst = IP4_ADDR_FULL_MASK;
+ }
+ if (spec.match_flags & EFX_FILTER_MATCH_REM_HOST) {
+ ip_entry->ip4src = spec.rem_host[0];
+ ip_mask->ip4src = IP4_ADDR_FULL_MASK;
+ }
+ if (spec.match_flags & EFX_FILTER_MATCH_LOC_PORT) {
+ ip_entry->pdst = spec.loc_port;
+ ip_mask->pdst = PORT_FULL_MASK;
+ }
+ if (spec.match_flags & EFX_FILTER_MATCH_REM_PORT) {
+ ip_entry->psrc = spec.rem_port;
+ ip_mask->psrc = PORT_FULL_MASK;
+ }
+ } else if (!(spec.match_flags &
+ ~(EFX_FILTER_MATCH_LOC_MAC | EFX_FILTER_MATCH_LOC_MAC_IG |
+ EFX_FILTER_MATCH_REM_MAC | EFX_FILTER_MATCH_ETHER_TYPE |
+ EFX_FILTER_MATCH_OUTER_VID))) {
rule->flow_type = ETHER_FLOW;
- memset(mac_mask->h_dest, ~0, ETH_ALEN);
- if (vid != EFX_FILTER_VID_UNSPEC) {
- rule->flow_type |= FLOW_EXT;
- rule->h_ext.vlan_tci = htons(vid);
- rule->m_ext.vlan_tci = htons(0xfff);
+ if (spec.match_flags &
+ (EFX_FILTER_MATCH_LOC_MAC | EFX_FILTER_MATCH_LOC_MAC_IG)) {
+ memcpy(mac_entry->h_dest, spec.loc_mac, ETH_ALEN);
+ if (spec.match_flags & EFX_FILTER_MATCH_LOC_MAC)
+ memset(mac_mask->h_dest, ~0, ETH_ALEN);
+ else
+ memcpy(mac_mask->h_dest, mac_addr_ig_mask,
+ ETH_ALEN);
}
- return 0;
+ if (spec.match_flags & EFX_FILTER_MATCH_REM_MAC) {
+ memcpy(mac_entry->h_source, spec.rem_mac, ETH_ALEN);
+ memset(mac_mask->h_source, ~0, ETH_ALEN);
+ }
+ if (spec.match_flags & EFX_FILTER_MATCH_ETHER_TYPE) {
+ mac_entry->h_proto = spec.ether_type;
+ mac_mask->h_proto = ETHER_TYPE_FULL_MASK;
+ }
+ } else {
+ /* The above should handle all filters that we insert */
+ WARN_ON(1);
+ return -EINVAL;
}
- rc = efx_filter_get_ipv4_local(&spec, &proto,
- &ip_entry->ip4dst, &ip_entry->pdst);
- if (rc != 0) {
- rc = efx_filter_get_ipv4_full(
- &spec, &proto, &ip_entry->ip4dst, &ip_entry->pdst,
- &ip_entry->ip4src, &ip_entry->psrc);
- EFX_WARN_ON_PARANOID(rc);
- ip_mask->ip4src = IP4_ADDR_FULL_MASK;
- ip_mask->psrc = PORT_FULL_MASK;
+ if (spec.match_flags & EFX_FILTER_MATCH_OUTER_VID) {
+ rule->flow_type |= FLOW_EXT;
+ rule->h_ext.vlan_tci = spec.outer_vid;
+ rule->m_ext.vlan_tci = htons(0xfff);
}
- rule->flow_type = (proto == IPPROTO_TCP) ? TCP_V4_FLOW : UDP_V4_FLOW;
- ip_mask->ip4dst = IP4_ADDR_FULL_MASK;
- ip_mask->pdst = PORT_FULL_MASK;
+
return rc;
}
@@ -969,80 +996,78 @@ static int efx_ethtool_set_class_rule(struct efx_nic *efx,
(rule->ring_cookie == RX_CLS_FLOW_DISC) ?
EFX_FILTER_RX_DMAQ_ID_DROP : rule->ring_cookie);
- switch (rule->flow_type) {
+ switch (rule->flow_type & ~FLOW_EXT) {
case TCP_V4_FLOW:
- case UDP_V4_FLOW: {
- u8 proto = (rule->flow_type == TCP_V4_FLOW ?
- IPPROTO_TCP : IPPROTO_UDP);
-
- /* Must match all of destination, */
- if (!(ip_mask->ip4dst == IP4_ADDR_FULL_MASK &&
- ip_mask->pdst == PORT_FULL_MASK))
- return -EINVAL;
- /* all or none of source, */
- if ((ip_mask->ip4src || ip_mask->psrc) &&
- !(ip_mask->ip4src == IP4_ADDR_FULL_MASK &&
- ip_mask->psrc == PORT_FULL_MASK))
- return -EINVAL;
- /* and nothing else */
- if (ip_mask->tos || rule->m_ext.vlan_tci)
+ case UDP_V4_FLOW:
+ spec.match_flags = (EFX_FILTER_MATCH_ETHER_TYPE |
+ EFX_FILTER_MATCH_IP_PROTO);
+ spec.ether_type = htons(ETH_P_IP);
+ spec.ip_proto = ((rule->flow_type & ~FLOW_EXT) == TCP_V4_FLOW ?
+ IPPROTO_TCP : IPPROTO_UDP);
+ if (ip_mask->ip4dst) {
+ if (ip_mask->ip4dst != IP4_ADDR_FULL_MASK)
+ return -EINVAL;
+ spec.match_flags |= EFX_FILTER_MATCH_LOC_HOST;
+ spec.loc_host[0] = ip_entry->ip4dst;
+ }
+ if (ip_mask->ip4src) {
+ if (ip_mask->ip4src != IP4_ADDR_FULL_MASK)
+ return -EINVAL;
+ spec.match_flags |= EFX_FILTER_MATCH_REM_HOST;
+ spec.rem_host[0] = ip_entry->ip4src;
+ }
+ if (ip_mask->pdst) {
+ if (ip_mask->pdst != PORT_FULL_MASK)
+ return -EINVAL;
+ spec.match_flags |= EFX_FILTER_MATCH_LOC_PORT;
+ spec.loc_port = ip_entry->pdst;
+ }
+ if (ip_mask->psrc) {
+ if (ip_mask->psrc != PORT_FULL_MASK)
+ return -EINVAL;
+ spec.match_flags |= EFX_FILTER_MATCH_REM_PORT;
+ spec.rem_port = ip_entry->psrc;
+ }
+ if (ip_mask->tos)
return -EINVAL;
-
- if (ip_mask->ip4src)
- rc = efx_filter_set_ipv4_full(&spec, proto,
- ip_entry->ip4dst,
- ip_entry->pdst,
- ip_entry->ip4src,
- ip_entry->psrc);
- else
- rc = efx_filter_set_ipv4_local(&spec, proto,
- ip_entry->ip4dst,
- ip_entry->pdst);
- if (rc)
- return rc;
break;
- }
-
- case ETHER_FLOW | FLOW_EXT:
- case ETHER_FLOW: {
- u16 vlan_tag_mask = (rule->flow_type & FLOW_EXT ?
- ntohs(rule->m_ext.vlan_tci) : 0);
- /* Must not match on source address or Ethertype */
- if (!is_zero_ether_addr(mac_mask->h_source) ||
- mac_mask->h_proto)
- return -EINVAL;
-
- /* Is it a default UC or MC filter? */
- if (ether_addr_equal(mac_mask->h_dest, mac_addr_mc_mask) &&
- vlan_tag_mask == 0) {
- if (is_multicast_ether_addr(mac_entry->h_dest))
- rc = efx_filter_set_mc_def(&spec);
+ case ETHER_FLOW:
+ if (!is_zero_ether_addr(mac_mask->h_dest)) {
+ if (ether_addr_equal(mac_mask->h_dest,
+ mac_addr_ig_mask))
+ spec.match_flags |= EFX_FILTER_MATCH_LOC_MAC_IG;
+ else if (is_broadcast_ether_addr(mac_mask->h_dest))
+ spec.match_flags |= EFX_FILTER_MATCH_LOC_MAC;
else
- rc = efx_filter_set_uc_def(&spec);
+ return -EINVAL;
+ memcpy(spec.loc_mac, mac_entry->h_dest, ETH_ALEN);
}
- /* Otherwise, it must match all of destination and all
- * or none of VID.
- */
- else if (is_broadcast_ether_addr(mac_mask->h_dest) &&
- (vlan_tag_mask == 0xfff || vlan_tag_mask == 0)) {
- rc = efx_filter_set_eth_local(
- &spec,
- vlan_tag_mask ?
- ntohs(rule->h_ext.vlan_tci) : EFX_FILTER_VID_UNSPEC,
- mac_entry->h_dest);
- } else {
- rc = -EINVAL;
+ if (!is_zero_ether_addr(mac_mask->h_source)) {
+ if (!is_broadcast_ether_addr(mac_mask->h_source))
+ return -EINVAL;
+ spec.match_flags |= EFX_FILTER_MATCH_REM_MAC;
+ memcpy(spec.rem_mac, mac_entry->h_source, ETH_ALEN);
+ }
+ if (mac_mask->h_proto) {
+ if (mac_mask->h_proto != ETHER_TYPE_FULL_MASK)
+ return -EINVAL;
+ spec.match_flags |= EFX_FILTER_MATCH_ETHER_TYPE;
+ spec.ether_type = mac_entry->h_proto;
}
- if (rc)
- return rc;
break;
- }
default:
return -EINVAL;
}
+ if ((rule->flow_type & FLOW_EXT) && rule->m_ext.vlan_tci) {
+ if (rule->m_ext.vlan_tci != htons(0xfff))
+ return -EINVAL;
+ spec.match_flags |= EFX_FILTER_MATCH_OUTER_VID;
+ spec.outer_vid = rule->h_ext.vlan_tci;
+ }
+
rc = efx_filter_insert_filter(efx, &spec, true);
if (rc < 0)
return rc;
diff --git a/drivers/net/ethernet/sfc/filter.c b/drivers/net/ethernet/sfc/filter.c
index c547630..96f8f2e 100644
--- a/drivers/net/ethernet/sfc/filter.c
+++ b/drivers/net/ethernet/sfc/filter.c
@@ -32,6 +32,18 @@
* counter-productive. */
#define EFX_FARCH_FILTER_CTL_SRCH_HINT_MAX 5
+enum efx_farch_filter_type {
+ EFX_FARCH_FILTER_TCP_FULL = 0,
+ EFX_FARCH_FILTER_TCP_WILD,
+ EFX_FARCH_FILTER_UDP_FULL,
+ EFX_FARCH_FILTER_UDP_WILD,
+ EFX_FARCH_FILTER_MAC_FULL = 4,
+ EFX_FARCH_FILTER_MAC_WILD,
+ EFX_FARCH_FILTER_UC_DEF = 8,
+ EFX_FARCH_FILTER_MC_DEF,
+ EFX_FARCH_FILTER_TYPE_COUNT, /* number of specific types */
+};
+
enum efx_farch_filter_table_id {
EFX_FARCH_FILTER_TABLE_RX_IP = 0,
EFX_FARCH_FILTER_TABLE_RX_MAC,
@@ -62,7 +74,7 @@ struct efx_farch_filter_table {
unsigned used; /* number currently used */
unsigned long *used_bitmap;
struct efx_farch_filter_spec *spec;
- unsigned search_depth[EFX_FILTER_TYPE_COUNT];
+ unsigned search_depth[EFX_FARCH_FILTER_TYPE_COUNT];
};
struct efx_filter_state {
@@ -106,33 +118,22 @@ static enum efx_farch_filter_table_id
efx_farch_filter_spec_table_id(const struct efx_farch_filter_spec *spec)
{
BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
- (EFX_FILTER_TCP_FULL >> 2));
+ (EFX_FARCH_FILTER_TCP_FULL >> 2));
BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
- (EFX_FILTER_TCP_WILD >> 2));
+ (EFX_FARCH_FILTER_TCP_WILD >> 2));
BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
- (EFX_FILTER_UDP_FULL >> 2));
+ (EFX_FARCH_FILTER_UDP_FULL >> 2));
BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
- (EFX_FILTER_UDP_WILD >> 2));
+ (EFX_FARCH_FILTER_UDP_WILD >> 2));
BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_MAC !=
- (EFX_FILTER_MAC_FULL >> 2));
+ (EFX_FARCH_FILTER_MAC_FULL >> 2));
BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_MAC !=
- (EFX_FILTER_MAC_WILD >> 2));
+ (EFX_FARCH_FILTER_MAC_WILD >> 2));
BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_TX_MAC !=
EFX_FARCH_FILTER_TABLE_RX_MAC + 2);
- EFX_BUG_ON_PARANOID(spec->type == EFX_FILTER_UNSPEC);
return (spec->type >> 2) + ((spec->flags & EFX_FILTER_FLAG_TX) ? 2 : 0);
}
-static struct efx_farch_filter_table *
-efx_farch_filter_spec_table(struct efx_filter_state *state,
- const struct efx_farch_filter_spec *spec)
-{
- if (spec->type == EFX_FILTER_UNSPEC)
- return NULL;
- else
- return &state->table[efx_farch_filter_spec_table_id(spec)];
-}
-
static void
efx_farch_filter_table_reset_search_depth(struct efx_farch_filter_table *table)
{
@@ -149,27 +150,27 @@ static void efx_farch_filter_push_rx_config(struct efx_nic *efx)
table = &state->table[EFX_FARCH_FILTER_TABLE_RX_IP];
EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_TCP_FULL_SRCH_LIMIT,
- table->search_depth[EFX_FILTER_TCP_FULL] +
+ table->search_depth[EFX_FARCH_FILTER_TCP_FULL] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_TCP_WILD_SRCH_LIMIT,
- table->search_depth[EFX_FILTER_TCP_WILD] +
+ table->search_depth[EFX_FARCH_FILTER_TCP_WILD] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_UDP_FULL_SRCH_LIMIT,
- table->search_depth[EFX_FILTER_UDP_FULL] +
+ table->search_depth[EFX_FARCH_FILTER_UDP_FULL] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_UDP_WILD_SRCH_LIMIT,
- table->search_depth[EFX_FILTER_UDP_WILD] +
+ table->search_depth[EFX_FARCH_FILTER_UDP_WILD] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
table = &state->table[EFX_FARCH_FILTER_TABLE_RX_MAC];
if (table->size) {
EFX_SET_OWORD_FIELD(
filter_ctl, FRF_CZ_ETHERNET_FULL_SEARCH_LIMIT,
- table->search_depth[EFX_FILTER_MAC_FULL] +
+ table->search_depth[EFX_FARCH_FILTER_MAC_FULL] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
EFX_SET_OWORD_FIELD(
filter_ctl, FRF_CZ_ETHERNET_WILDCARD_SEARCH_LIMIT,
- table->search_depth[EFX_FILTER_MAC_WILD] +
+ table->search_depth[EFX_FARCH_FILTER_MAC_WILD] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
}
@@ -225,223 +226,202 @@ static void efx_farch_filter_push_tx_limits(struct efx_nic *efx)
if (table->size) {
EFX_SET_OWORD_FIELD(
tx_cfg, FRF_CZ_TX_ETH_FILTER_FULL_SEARCH_RANGE,
- table->search_depth[EFX_FILTER_MAC_FULL] +
+ table->search_depth[EFX_FARCH_FILTER_MAC_FULL] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
EFX_SET_OWORD_FIELD(
tx_cfg, FRF_CZ_TX_ETH_FILTER_WILD_SEARCH_RANGE,
- table->search_depth[EFX_FILTER_MAC_WILD] +
+ table->search_depth[EFX_FARCH_FILTER_MAC_WILD] +
EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
}
efx_writeo(efx, &tx_cfg, FR_AZ_TX_CFG);
}
-static inline void __efx_filter_set_ipv4(struct efx_filter_spec *spec,
- __be32 host1, __be16 port1,
- __be32 host2, __be16 port2)
-{
- spec->data[0] = ntohl(host1) << 16 | ntohs(port1);
- spec->data[1] = ntohs(port2) << 16 | ntohl(host1) >> 16;
- spec->data[2] = ntohl(host2);
-}
-
-static inline void __efx_filter_get_ipv4(const struct efx_filter_spec *spec,
- __be32 *host1, __be16 *port1,
- __be32 *host2, __be16 *port2)
-{
- *host1 = htonl(spec->data[0] >> 16 | spec->data[1] << 16);
- *port1 = htons(spec->data[0]);
- *host2 = htonl(spec->data[2]);
- *port2 = htons(spec->data[1] >> 16);
-}
-
-/**
- * efx_filter_set_ipv4_local - specify IPv4 host, transport protocol and port
- * @spec: Specification to initialise
- * @proto: Transport layer protocol number
- * @host: Local host address (network byte order)
- * @port: Local port (network byte order)
- */
-int efx_filter_set_ipv4_local(struct efx_filter_spec *spec, u8 proto,
- __be32 host, __be16 port)
+static int
+efx_farch_filter_from_gen_spec(struct efx_farch_filter_spec *spec,
+ const struct efx_filter_spec *gen_spec)
{
- __be32 host1;
- __be16 port1;
-
- EFX_BUG_ON_PARANOID(!(spec->flags & EFX_FILTER_FLAG_RX));
-
- /* This cannot currently be combined with other filtering */
- if (spec->type != EFX_FILTER_UNSPEC)
- return -EPROTONOSUPPORT;
+ bool is_full = false;
- if (port == 0)
+ if ((gen_spec->flags & EFX_FILTER_FLAG_RX_RSS) &&
+ gen_spec->rss_context != EFX_FILTER_RSS_CONTEXT_DEFAULT)
return -EINVAL;
- switch (proto) {
- case IPPROTO_TCP:
- spec->type = EFX_FILTER_TCP_WILD;
- break;
- case IPPROTO_UDP:
- spec->type = EFX_FILTER_UDP_WILD;
- break;
- default:
- return -EPROTONOSUPPORT;
- }
-
- /* Filter is constructed in terms of source and destination,
- * with the odd wrinkle that the ports are swapped in a UDP
- * wildcard filter. We need to convert from local and remote
- * (= zero for wildcard) addresses.
- */
- host1 = 0;
- if (proto != IPPROTO_UDP) {
- port1 = 0;
- } else {
- port1 = port;
- port = 0;
- }
-
- __efx_filter_set_ipv4(spec, host1, port1, host, port);
- return 0;
-}
+ spec->priority = gen_spec->priority;
+ spec->flags = gen_spec->flags;
+ spec->dmaq_id = gen_spec->dmaq_id;
-int efx_filter_get_ipv4_local(const struct efx_filter_spec *spec,
- u8 *proto, __be32 *host, __be16 *port)
-{
- __be32 host1;
- __be16 port1;
+ switch (gen_spec->match_flags) {
+ case (EFX_FILTER_MATCH_ETHER_TYPE | EFX_FILTER_MATCH_IP_PROTO |
+ EFX_FILTER_MATCH_LOC_HOST | EFX_FILTER_MATCH_LOC_PORT |
+ EFX_FILTER_MATCH_REM_HOST | EFX_FILTER_MATCH_REM_PORT):
+ is_full = true;
+ /* fall through */
+ case (EFX_FILTER_MATCH_ETHER_TYPE | EFX_FILTER_MATCH_IP_PROTO |
+ EFX_FILTER_MATCH_LOC_HOST | EFX_FILTER_MATCH_LOC_PORT): {
+ __be32 rhost, host1, host2;
+ __be16 rport, port1, port2;
- switch (spec->type) {
- case EFX_FILTER_TCP_WILD:
- *proto = IPPROTO_TCP;
- __efx_filter_get_ipv4(spec, &host1, &port1, host, port);
- return 0;
- case EFX_FILTER_UDP_WILD:
- *proto = IPPROTO_UDP;
- __efx_filter_get_ipv4(spec, &host1, port, host, &port1);
- return 0;
- default:
- return -EINVAL;
- }
-}
-
-/**
- * efx_filter_set_ipv4_full - specify IPv4 hosts, transport protocol and ports
- * @spec: Specification to initialise
- * @proto: Transport layer protocol number
- * @host: Local host address (network byte order)
- * @port: Local port (network byte order)
- * @rhost: Remote host address (network byte order)
- * @rport: Remote port (network byte order)
- */
-int efx_filter_set_ipv4_full(struct efx_filter_spec *spec, u8 proto,
- __be32 host, __be16 port,
- __be32 rhost, __be16 rport)
-{
- EFX_BUG_ON_PARANOID(!(spec->flags & EFX_FILTER_FLAG_RX));
+ EFX_BUG_ON_PARANOID(!(gen_spec->flags & EFX_FILTER_FLAG_RX));
- /* This cannot currently be combined with other filtering */
- if (spec->type != EFX_FILTER_UNSPEC)
- return -EPROTONOSUPPORT;
+ if (gen_spec->ether_type != htons(ETH_P_IP))
+ return -EPROTONOSUPPORT;
+ if (gen_spec->loc_port == 0 ||
+ (is_full && gen_spec->rem_port == 0))
+ return -EADDRNOTAVAIL;
+ switch (gen_spec->ip_proto) {
+ case IPPROTO_TCP:
+ spec->type = (is_full ? EFX_FARCH_FILTER_TCP_FULL :
+ EFX_FARCH_FILTER_TCP_WILD);
+ break;
+ case IPPROTO_UDP:
+ spec->type = (is_full ? EFX_FARCH_FILTER_UDP_FULL :
+ EFX_FARCH_FILTER_UDP_WILD);
+ break;
+ default:
+ return -EPROTONOSUPPORT;
+ }
- if (port == 0 || rport == 0)
- return -EINVAL;
+ /* Filter is constructed in terms of source and destination,
+ * with the odd wrinkle that the ports are swapped in a UDP
+ * wildcard filter. We need to convert from local and remote
+ * (= zero for wildcard) addresses.
+ */
+ rhost = is_full ? gen_spec->rem_host[0] : 0;
+ rport = is_full ? gen_spec->rem_port : 0;
+ host1 = rhost;
+ host2 = gen_spec->loc_host[0];
+ if (!is_full && gen_spec->ip_proto == IPPROTO_UDP) {
+ port1 = gen_spec->loc_port;
+ port2 = rport;
+ } else {
+ port1 = rport;
+ port2 = gen_spec->loc_port;
+ }
+ spec->data[0] = ntohl(host1) << 16 | ntohs(port1);
+ spec->data[1] = ntohs(port2) << 16 | ntohl(host1) >> 16;
+ spec->data[2] = ntohl(host2);
- switch (proto) {
- case IPPROTO_TCP:
- spec->type = EFX_FILTER_TCP_FULL;
break;
- case IPPROTO_UDP:
- spec->type = EFX_FILTER_UDP_FULL;
- break;
- default:
- return -EPROTONOSUPPORT;
}
- __efx_filter_set_ipv4(spec, rhost, rport, host, port);
- return 0;
-}
-
-int efx_filter_get_ipv4_full(const struct efx_filter_spec *spec,
- u8 *proto, __be32 *host, __be16 *port,
- __be32 *rhost, __be16 *rport)
-{
- switch (spec->type) {
- case EFX_FILTER_TCP_FULL:
- *proto = IPPROTO_TCP;
+ case EFX_FILTER_MATCH_LOC_MAC | EFX_FILTER_MATCH_OUTER_VID:
+ is_full = true;
+ /* fall through */
+ case EFX_FILTER_MATCH_LOC_MAC:
+ spec->type = (is_full ? EFX_FARCH_FILTER_MAC_FULL :
+ EFX_FARCH_FILTER_MAC_WILD);
+ spec->data[0] = is_full ? ntohs(gen_spec->outer_vid) : 0;
+ spec->data[1] = (gen_spec->loc_mac[2] << 24 |
+ gen_spec->loc_mac[3] << 16 |
+ gen_spec->loc_mac[4] << 8 |
+ gen_spec->loc_mac[5]);
+ spec->data[2] = (gen_spec->loc_mac[0] << 8 |
+ gen_spec->loc_mac[1]);
break;
- case EFX_FILTER_UDP_FULL:
- *proto = IPPROTO_UDP;
+
+ case EFX_FILTER_MATCH_LOC_MAC_IG:
+ spec->type = (is_multicast_ether_addr(gen_spec->loc_mac) ?
+ EFX_FARCH_FILTER_MC_DEF :
+ EFX_FARCH_FILTER_UC_DEF);
+ memset(spec->data, 0, sizeof(spec->data)); /* ensure equality */
break;
+
default:
- return -EINVAL;
+ return -EPROTONOSUPPORT;
}
- __efx_filter_get_ipv4(spec, rhost, rport, host, port);
return 0;
}
-/**
- * efx_filter_set_eth_local - specify local Ethernet address and optional VID
- * @spec: Specification to initialise
- * @vid: VLAN ID to match, or %EFX_FILTER_VID_UNSPEC
- * @addr: Local Ethernet MAC address
- */
-int efx_filter_set_eth_local(struct efx_filter_spec *spec,
- u16 vid, const u8 *addr)
+static void
+efx_farch_filter_to_gen_spec(struct efx_filter_spec *gen_spec,
+ const struct efx_farch_filter_spec *spec)
{
- EFX_BUG_ON_PARANOID(!(spec->flags &
- (EFX_FILTER_FLAG_RX | EFX_FILTER_FLAG_TX)));
-
- /* This cannot currently be combined with other filtering */
- if (spec->type != EFX_FILTER_UNSPEC)
- return -EPROTONOSUPPORT;
-
- if (vid == EFX_FILTER_VID_UNSPEC) {
- spec->type = EFX_FILTER_MAC_WILD;
- spec->data[0] = 0;
- } else {
- spec->type = EFX_FILTER_MAC_FULL;
- spec->data[0] = vid;
- }
+ bool is_full = false;
- spec->data[1] = addr[2] << 24 | addr[3] << 16 | addr[4] << 8 | addr[5];
- spec->data[2] = addr[0] << 8 | addr[1];
- return 0;
-}
+ /* *gen_spec should be completely initialised, to be consistent
+ * with efx_filter_init_{rx,tx}() and in case we want to copy
+ * it back to userland.
+ */
+ memset(gen_spec, 0, sizeof(*gen_spec));
-/**
- * efx_filter_set_uc_def - specify matching otherwise-unmatched unicast
- * @spec: Specification to initialise
- */
-int efx_filter_set_uc_def(struct efx_filter_spec *spec)
-{
- EFX_BUG_ON_PARANOID(!(spec->flags &
- (EFX_FILTER_FLAG_RX | EFX_FILTER_FLAG_TX)));
+ gen_spec->priority = spec->priority;
+ gen_spec->flags = spec->flags;
+ gen_spec->dmaq_id = spec->dmaq_id;
- if (spec->type != EFX_FILTER_UNSPEC)
- return -EINVAL;
+ switch (spec->type) {
+ case EFX_FARCH_FILTER_TCP_FULL:
+ case EFX_FARCH_FILTER_UDP_FULL:
+ is_full = true;
+ /* fall through */
+ case EFX_FARCH_FILTER_TCP_WILD:
+ case EFX_FARCH_FILTER_UDP_WILD: {
+ __be32 host1, host2;
+ __be16 port1, port2;
+
+ gen_spec->match_flags =
+ EFX_FILTER_MATCH_ETHER_TYPE |
+ EFX_FILTER_MATCH_IP_PROTO |
+ EFX_FILTER_MATCH_LOC_HOST | EFX_FILTER_MATCH_LOC_PORT;
+ if (is_full)
+ gen_spec->match_flags |= (EFX_FILTER_MATCH_REM_HOST |
+ EFX_FILTER_MATCH_REM_PORT);
+ gen_spec->ether_type = htons(ETH_P_IP);
+ gen_spec->ip_proto =
+ (spec->type == EFX_FARCH_FILTER_TCP_FULL ||
+ spec->type == EFX_FARCH_FILTER_TCP_WILD) ?
+ IPPROTO_TCP : IPPROTO_UDP;
+
+ host1 = htonl(spec->data[0] >> 16 | spec->data[1] << 16);
+ port1 = htons(spec->data[0]);
+ host2 = htonl(spec->data[2]);
+ port2 = htons(spec->data[1] >> 16);
+ if (spec->flags & EFX_FILTER_FLAG_TX) {
+ gen_spec->loc_host[0] = host1;
+ gen_spec->rem_host[0] = host2;
+ } else {
+ gen_spec->loc_host[0] = host2;
+ gen_spec->rem_host[0] = host1;
+ }
+ if (!!(gen_spec->flags & EFX_FILTER_FLAG_TX) ^
+ (!is_full && gen_spec->ip_proto == IPPROTO_UDP)) {
+ gen_spec->loc_port = port1;
+ gen_spec->rem_port = port2;
+ } else {
+ gen_spec->loc_port = port2;
+ gen_spec->rem_port = port1;
+ }
- spec->type = EFX_FILTER_UC_DEF;
- memset(spec->data, 0, sizeof(spec->data)); /* ensure equality */
- return 0;
-}
+ break;
+ }
-/**
- * efx_filter_set_mc_def - specify matching otherwise-unmatched multicast
- * @spec: Specification to initialise
- */
-int efx_filter_set_mc_def(struct efx_filter_spec *spec)
-{
- EFX_BUG_ON_PARANOID(!(spec->flags &
- (EFX_FILTER_FLAG_RX | EFX_FILTER_FLAG_TX)));
+ case EFX_FARCH_FILTER_MAC_FULL:
+ is_full = true;
+ /* fall through */
+ case EFX_FARCH_FILTER_MAC_WILD:
+ gen_spec->match_flags = EFX_FILTER_MATCH_LOC_MAC;
+ if (is_full)
+ gen_spec->match_flags |= EFX_FILTER_MATCH_OUTER_VID;
+ gen_spec->loc_mac[0] = spec->data[2] >> 8;
+ gen_spec->loc_mac[1] = spec->data[2];
+ gen_spec->loc_mac[2] = spec->data[1] >> 24;
+ gen_spec->loc_mac[3] = spec->data[1] >> 16;
+ gen_spec->loc_mac[4] = spec->data[1] >> 8;
+ gen_spec->loc_mac[5] = spec->data[1];
+ gen_spec->outer_vid = htons(spec->data[0]);
+ break;
- if (spec->type != EFX_FILTER_UNSPEC)
- return -EINVAL;
+ case EFX_FARCH_FILTER_UC_DEF:
+ case EFX_FARCH_FILTER_MC_DEF:
+ gen_spec->match_flags = EFX_FILTER_MATCH_LOC_MAC_IG;
+ gen_spec->loc_mac[0] = spec->type == EFX_FARCH_FILTER_MC_DEF;
+ break;
- spec->type = EFX_FILTER_MC_DEF;
- memset(spec->data, 0, sizeof(spec->data)); /* ensure equality */
- return 0;
+ default:
+ WARN_ON(1);
+ break;
+ }
}
static void
@@ -455,7 +435,7 @@ efx_farch_filter_reset_rx_def(struct efx_nic *efx, unsigned filter_idx)
/* If there's only one channel then disable RSS for non VF
* traffic, thereby allowing VFs to use RSS when the PF can't.
*/
- spec->type = EFX_FILTER_UC_DEF + filter_idx;
+ spec->type = EFX_FARCH_FILTER_UC_DEF + filter_idx;
spec->priority = EFX_FILTER_PRI_MANUAL;
spec->flags = (EFX_FILTER_FLAG_RX |
(efx->n_rx_channels > 1 ? EFX_FILTER_FLAG_RX_RSS : 0) |
@@ -464,29 +444,6 @@ efx_farch_filter_reset_rx_def(struct efx_nic *efx, unsigned filter_idx)
table->used_bitmap[0] |= 1 << filter_idx;
}
-int efx_filter_get_eth_local(const struct efx_filter_spec *spec,
- u16 *vid, u8 *addr)
-{
- switch (spec->type) {
- case EFX_FILTER_MAC_WILD:
- *vid = EFX_FILTER_VID_UNSPEC;
- break;
- case EFX_FILTER_MAC_FULL:
- *vid = spec->data[0];
- break;
- default:
- return -EINVAL;
- }
-
- addr[0] = spec->data[2] >> 8;
- addr[1] = spec->data[2];
- addr[2] = spec->data[1] >> 24;
- addr[3] = spec->data[1] >> 16;
- addr[4] = spec->data[1] >> 8;
- addr[5] = spec->data[1];
- return 0;
-}
-
/* Build a filter entry and return its n-tuple key. */
static u32 efx_farch_filter_build(efx_oword_t *filter,
struct efx_farch_filter_spec *spec)
@@ -495,8 +452,8 @@ static u32 efx_farch_filter_build(efx_oword_t *filter,
switch (efx_farch_filter_spec_table_id(spec)) {
case EFX_FARCH_FILTER_TABLE_RX_IP: {
- bool is_udp = (spec->type == EFX_FILTER_UDP_FULL ||
- spec->type == EFX_FILTER_UDP_WILD);
+ bool is_udp = (spec->type == EFX_FARCH_FILTER_UDP_FULL ||
+ spec->type == EFX_FARCH_FILTER_UDP_WILD);
EFX_POPULATE_OWORD_7(
*filter,
FRF_BZ_RSS_EN,
@@ -513,7 +470,7 @@ static u32 efx_farch_filter_build(efx_oword_t *filter,
}
case EFX_FARCH_FILTER_TABLE_RX_MAC: {
- bool is_wild = spec->type == EFX_FILTER_MAC_WILD;
+ bool is_wild = spec->type == EFX_FARCH_FILTER_MAC_WILD;
EFX_POPULATE_OWORD_7(
*filter,
FRF_CZ_RMFT_RSS_EN,
@@ -530,7 +487,7 @@ static u32 efx_farch_filter_build(efx_oword_t *filter,
}
case EFX_FARCH_FILTER_TABLE_TX_MAC: {
- bool is_wild = spec->type == EFX_FILTER_MAC_WILD;
+ bool is_wild = spec->type == EFX_FARCH_FILTER_MAC_WILD;
EFX_POPULATE_OWORD_5(*filter,
FRF_CZ_TMFT_TXQ_ID, spec->dmaq_id,
FRF_CZ_TMFT_WILDCARD_MATCH, is_wild,
@@ -573,15 +530,15 @@ static bool efx_farch_filter_equal(const struct efx_farch_filter_spec *left,
#define EFX_FARCH_FILTER_MATCH_PRI_COUNT 5
-static const u8 efx_farch_filter_type_match_pri[EFX_FILTER_TYPE_COUNT] = {
- [EFX_FILTER_TCP_FULL] = 0,
- [EFX_FILTER_UDP_FULL] = 0,
- [EFX_FILTER_TCP_WILD] = 1,
- [EFX_FILTER_UDP_WILD] = 1,
- [EFX_FILTER_MAC_FULL] = 2,
- [EFX_FILTER_MAC_WILD] = 3,
- [EFX_FILTER_UC_DEF] = 4,
- [EFX_FILTER_MC_DEF] = 4,
+static const u8 efx_farch_filter_type_match_pri[EFX_FARCH_FILTER_TYPE_COUNT] = {
+ [EFX_FARCH_FILTER_TCP_FULL] = 0,
+ [EFX_FARCH_FILTER_UDP_FULL] = 0,
+ [EFX_FARCH_FILTER_TCP_WILD] = 1,
+ [EFX_FARCH_FILTER_UDP_WILD] = 1,
+ [EFX_FARCH_FILTER_MAC_FULL] = 2,
+ [EFX_FARCH_FILTER_MAC_WILD] = 3,
+ [EFX_FARCH_FILTER_UC_DEF] = 4,
+ [EFX_FARCH_FILTER_MC_DEF] = 4,
};
static const enum efx_farch_filter_table_id efx_farch_filter_range_table[] = {
@@ -672,11 +629,12 @@ s32 efx_filter_insert_filter(struct efx_nic *efx,
unsigned int depth = 0;
int rc;
- /* XXX efx_farch_filter_spec and efx_filter_spec will diverge in future */
- memcpy(&spec, gen_spec, sizeof(*gen_spec));
+ rc = efx_farch_filter_from_gen_spec(&spec, gen_spec);
+ if (rc)
+ return rc;
- table = efx_farch_filter_spec_table(state, &spec);
- if (!table || table->size == 0)
+ table = &state->table[efx_farch_filter_spec_table_id(&spec)];
+ if (table->size == 0)
return -EINVAL;
netif_vdbg(efx, hw, efx->net_dev,
@@ -687,8 +645,8 @@ s32 efx_filter_insert_filter(struct efx_nic *efx,
/* One filter spec per type */
BUILD_BUG_ON(EFX_FARCH_FILTER_INDEX_UC_DEF != 0);
BUILD_BUG_ON(EFX_FARCH_FILTER_INDEX_MC_DEF !=
- EFX_FILTER_MC_DEF - EFX_FILTER_UC_DEF);
- rep_index = spec.type - EFX_FILTER_UC_DEF;
+ EFX_FARCH_FILTER_MC_DEF - EFX_FARCH_FILTER_UC_DEF);
+ rep_index = spec.type - EFX_FARCH_FILTER_UC_DEF;
ins_index = rep_index;
spin_lock_bh(&state->lock);
@@ -911,8 +869,7 @@ int efx_filter_get_filter_safe(struct efx_nic *efx,
if (test_bit(filter_idx, table->used_bitmap) &&
spec->priority == priority) {
- /* XXX efx_farch_filter_spec and efx_filter_spec will diverge */
- memcpy(spec_buf, spec, sizeof(*spec));
+ efx_farch_filter_to_gen_spec(spec_buf, spec);
rc = 0;
} else {
rc = -ENOENT;
diff --git a/drivers/net/ethernet/sfc/filter.h b/drivers/net/ethernet/sfc/filter.h
index b1170d4..1b410de 100644
--- a/drivers/net/ethernet/sfc/filter.h
+++ b/drivers/net/ethernet/sfc/filter.h
@@ -11,32 +11,49 @@
#define EFX_FILTER_H
#include <linux/types.h>
+#include <linux/if_ether.h>
+#include <asm/byteorder.h>
/**
- * enum efx_filter_type - type of hardware filter
- * @EFX_FILTER_TCP_FULL: Matching TCP/IPv4 4-tuple
- * @EFX_FILTER_TCP_WILD: Matching TCP/IPv4 destination (host, port)
- * @EFX_FILTER_UDP_FULL: Matching UDP/IPv4 4-tuple
- * @EFX_FILTER_UDP_WILD: Matching UDP/IPv4 destination (host, port)
- * @EFX_FILTER_MAC_FULL: Matching Ethernet destination MAC address, VID
- * @EFX_FILTER_MAC_WILD: Matching Ethernet destination MAC address
- * @EFX_FILTER_UC_DEF: Matching all otherwise unmatched unicast
- * @EFX_FILTER_MC_DEF: Matching all otherwise unmatched multicast
- * @EFX_FILTER_UNSPEC: Match type is unspecified
+ * enum efx_filter_match_flags - Flags for hardware filter match type
+ * @EFX_FILTER_MATCH_REM_HOST: Match by remote IP host address
+ * @EFX_FILTER_MATCH_LOC_HOST: Match by local IP host address
+ * @EFX_FILTER_MATCH_REM_MAC: Match by remote MAC address
+ * @EFX_FILTER_MATCH_REM_PORT: Match by remote TCP/UDP port
+ * @EFX_FILTER_MATCH_LOC_MAC: Match by local MAC address
+ * @EFX_FILTER_MATCH_LOC_PORT: Match by local TCP/UDP port
+ * @EFX_FILTER_MATCH_ETHER_TYPE: Match by Ether-type
+ * @EFX_FILTER_MATCH_INNER_VID: Match by inner VLAN ID
+ * @EFX_FILTER_MATCH_OUTER_VID: Match by outer VLAN ID
+ * @EFX_FILTER_MATCH_IP_PROTO: Match by IP transport protocol
+ * @EFX_FILTER_MATCH_LOC_MAC_IG: Match by local MAC address I/G bit.
+ * Used for RX default unicast and multicast/broadcast filters.
*
- * Falcon NICs only support the TCP/IPv4 and UDP/IPv4 filter types.
+ * Only some combinations are supported, depending on NIC type:
+ *
+ * - Falcon supports RX filters matching by {TCP,UDP}/IPv4 4-tuple or
+ * local 2-tuple (only implemented for Falcon B0)
+ *
+ * - Siena supports RX and TX filters matching by {TCP,UDP}/IPv4 4-tuple
+ * or local 2-tuple, or local MAC with or without outer VID, and RX
+ * default filters
+ *
+ * - Huntington supports filter matching controlled by firmware, potentially
+ * using {TCP,UDP}/IPv{4,6} 4-tuple or local 2-tuple, local MAC or I/G bit,
+ * with or without outer and inner VID
*/
-enum efx_filter_type {
- EFX_FILTER_TCP_FULL = 0,
- EFX_FILTER_TCP_WILD,
- EFX_FILTER_UDP_FULL,
- EFX_FILTER_UDP_WILD,
- EFX_FILTER_MAC_FULL = 4,
- EFX_FILTER_MAC_WILD,
- EFX_FILTER_UC_DEF = 8,
- EFX_FILTER_MC_DEF,
- EFX_FILTER_TYPE_COUNT, /* number of specific types */
- EFX_FILTER_UNSPEC = 0xf,
+enum efx_filter_match_flags {
+ EFX_FILTER_MATCH_REM_HOST = 0x0001,
+ EFX_FILTER_MATCH_LOC_HOST = 0x0002,
+ EFX_FILTER_MATCH_REM_MAC = 0x0004,
+ EFX_FILTER_MATCH_REM_PORT = 0x0008,
+ EFX_FILTER_MATCH_LOC_MAC = 0x0010,
+ EFX_FILTER_MATCH_LOC_PORT = 0x0020,
+ EFX_FILTER_MATCH_ETHER_TYPE = 0x0040,
+ EFX_FILTER_MATCH_INNER_VID = 0x0080,
+ EFX_FILTER_MATCH_OUTER_VID = 0x0100,
+ EFX_FILTER_MATCH_IP_PROTO = 0x0200,
+ EFX_FILTER_MATCH_LOC_MAC_IG = 0x0400,
};
/**
@@ -73,29 +90,55 @@ enum efx_filter_flags {
/**
* struct efx_filter_spec - specification for a hardware filter
- * @type: Type of match to be performed, from &enum efx_filter_type
+ * @match_flags: Match type flags, from &enum efx_filter_match_flags
* @priority: Priority of the filter, from &enum efx_filter_priority
* @flags: Miscellaneous flags, from &enum efx_filter_flags
+ * @rss_context: RSS context to use, if %EFX_FILTER_FLAG_RX_RSS is set
* @dmaq_id: Source/target queue index, or %EFX_FILTER_RX_DMAQ_ID_DROP for
* an RX drop filter
- * @data: Match data (type-dependent)
+ * @outer_vid: Outer VLAN ID to match, if %EFX_FILTER_MATCH_OUTER_VID is set
+ * @inner_vid: Inner VLAN ID to match, if %EFX_FILTER_MATCH_INNER_VID is set
+ * @loc_mac: Local MAC address to match, if %EFX_FILTER_MATCH_LOC_MAC or
+ * %EFX_FILTER_MATCH_LOC_MAC_IG is set
+ * @rem_mac: Remote MAC address to match, if %EFX_FILTER_MATCH_REM_MAC is set
+ * @ether_type: Ether-type to match, if %EFX_FILTER_MATCH_ETHER_TYPE is set
+ * @ip_proto: IP transport protocol to match, if %EFX_FILTER_MATCH_IP_PROTO
+ * is set
+ * @loc_host: Local IP host to match, if %EFX_FILTER_MATCH_LOC_HOST is set
+ * @rem_host: Remote IP host to match, if %EFX_FILTER_MATCH_REM_HOST is set
+ * @loc_port: Local TCP/UDP port to match, if %EFX_FILTER_MATCH_LOC_PORT is set
+ * @rem_port: Remote TCP/UDP port to match, if %EFX_FILTER_MATCH_REM_PORT is set
*
- * Use the efx_filter_set_*() functions to initialise the @type and
- * @data fields.
+ * The efx_filter_init_rx() or efx_filter_init_tx() function *must* be
+ * used to initialise the structure. The efx_filter_set_*() functions
+ * may then be used to set @rss_context, @match_flags and related
+ * fields.
*
* The @priority field is used by software to determine whether a new
* filter may replace an old one. The hardware priority of a filter
- * depends on the filter type.
+ * depends on which fields are matched.
*/
struct efx_filter_spec {
- u8 type:4;
- u8 priority:4;
- u8 flags;
- u16 dmaq_id;
- u32 data[3];
+ u32 match_flags:12;
+ u32 priority:2;
+ u32 flags:6;
+ u32 dmaq_id:12;
+ u32 rss_context;
+ __be16 outer_vid __aligned(4); /* allow jhash2() of match values */
+ __be16 inner_vid;
+ u8 loc_mac[ETH_ALEN];
+ u8 rem_mac[ETH_ALEN];
+ __be16 ether_type;
+ u8 ip_proto;
+ __be32 loc_host[4];
+ __be32 rem_host[4];
+ __be16 loc_port;
+ __be16 rem_port;
+ /* total 64 bytes */
};
enum {
+ EFX_FILTER_RSS_CONTEXT_DEFAULT = 0xffffffff,
EFX_FILTER_RX_DMAQ_ID_DROP = 0xfff
};
@@ -104,39 +147,116 @@ static inline void efx_filter_init_rx(struct efx_filter_spec *spec,
enum efx_filter_flags flags,
unsigned rxq_id)
{
- spec->type = EFX_FILTER_UNSPEC;
+ memset(spec, 0, sizeof(*spec));
spec->priority = priority;
spec->flags = EFX_FILTER_FLAG_RX | flags;
+ spec->rss_context = EFX_FILTER_RSS_CONTEXT_DEFAULT;
spec->dmaq_id = rxq_id;
}
static inline void efx_filter_init_tx(struct efx_filter_spec *spec,
unsigned txq_id)
{
- spec->type = EFX_FILTER_UNSPEC;
+ memset(spec, 0, sizeof(*spec));
spec->priority = EFX_FILTER_PRI_REQUIRED;
spec->flags = EFX_FILTER_FLAG_TX;
spec->dmaq_id = txq_id;
}
-extern int efx_filter_set_ipv4_local(struct efx_filter_spec *spec, u8 proto,
- __be32 host, __be16 port);
-extern int efx_filter_get_ipv4_local(const struct efx_filter_spec *spec,
- u8 *proto, __be32 *host, __be16 *port);
-extern int efx_filter_set_ipv4_full(struct efx_filter_spec *spec, u8 proto,
- __be32 host, __be16 port,
- __be32 rhost, __be16 rport);
-extern int efx_filter_get_ipv4_full(const struct efx_filter_spec *spec,
- u8 *proto, __be32 *host, __be16 *port,
- __be32 *rhost, __be16 *rport);
-extern int efx_filter_set_eth_local(struct efx_filter_spec *spec,
- u16 vid, const u8 *addr);
-extern int efx_filter_get_eth_local(const struct efx_filter_spec *spec,
- u16 *vid, u8 *addr);
-extern int efx_filter_set_uc_def(struct efx_filter_spec *spec);
-extern int efx_filter_set_mc_def(struct efx_filter_spec *spec);
+/**
+ * efx_filter_set_ipv4_local - specify IPv4 host, transport protocol and port
+ * @spec: Specification to initialise
+ * @proto: Transport layer protocol number
+ * @host: Local host address (network byte order)
+ * @port: Local port (network byte order)
+ */
+static inline int
+efx_filter_set_ipv4_local(struct efx_filter_spec *spec, u8 proto,
+ __be32 host, __be16 port)
+{
+ spec->match_flags |=
+ EFX_FILTER_MATCH_ETHER_TYPE | EFX_FILTER_MATCH_IP_PROTO |
+ EFX_FILTER_MATCH_LOC_HOST | EFX_FILTER_MATCH_LOC_PORT;
+ spec->ether_type = htons(ETH_P_IP);
+ spec->ip_proto = proto;
+ spec->loc_host[0] = host;
+ spec->loc_port = port;
+ return 0;
+}
+
+/**
+ * efx_filter_set_ipv4_full - specify IPv4 hosts, transport protocol and ports
+ * @spec: Specification to initialise
+ * @proto: Transport layer protocol number
+ * @lhost: Local host address (network byte order)
+ * @lport: Local port (network byte order)
+ * @rhost: Remote host address (network byte order)
+ * @rport: Remote port (network byte order)
+ */
+static inline int
+efx_filter_set_ipv4_full(struct efx_filter_spec *spec, u8 proto,
+ __be32 lhost, __be16 lport,
+ __be32 rhost, __be16 rport)
+{
+ spec->match_flags |=
+ EFX_FILTER_MATCH_ETHER_TYPE | EFX_FILTER_MATCH_IP_PROTO |
+ EFX_FILTER_MATCH_LOC_HOST | EFX_FILTER_MATCH_LOC_PORT |
+ EFX_FILTER_MATCH_REM_HOST | EFX_FILTER_MATCH_REM_PORT;
+ spec->ether_type = htons(ETH_P_IP);
+ spec->ip_proto = proto;
+ spec->loc_host[0] = lhost;
+ spec->loc_port = lport;
+ spec->rem_host[0] = rhost;
+ spec->rem_port = rport;
+ return 0;
+}
+
enum {
EFX_FILTER_VID_UNSPEC = 0xffff,
};
+/**
+ * efx_filter_set_eth_local - specify local Ethernet address and/or VID
+ * @spec: Specification to initialise
+ * @vid: Outer VLAN ID to match, or %EFX_FILTER_VID_UNSPEC
+ * @addr: Local Ethernet MAC address, or %NULL
+ */
+static inline int efx_filter_set_eth_local(struct efx_filter_spec *spec,
+ u16 vid, const u8 *addr)
+{
+ if (vid == EFX_FILTER_VID_UNSPEC && addr == NULL)
+ return -EINVAL;
+
+ if (vid != EFX_FILTER_VID_UNSPEC) {
+ spec->match_flags |= EFX_FILTER_MATCH_OUTER_VID;
+ spec->outer_vid = htons(vid);
+ }
+ if (addr != NULL) {
+ spec->match_flags |= EFX_FILTER_MATCH_LOC_MAC;
+ memcpy(spec->loc_mac, addr, ETH_ALEN);
+ }
+ return 0;
+}
+
+/**
+ * efx_filter_set_uc_def - specify matching otherwise-unmatched unicast
+ * @spec: Specification to initialise
+ */
+static inline int efx_filter_set_uc_def(struct efx_filter_spec *spec)
+{
+ spec->match_flags |= EFX_FILTER_MATCH_LOC_MAC_IG;
+ return 0;
+}
+
+/**
+ * efx_filter_set_mc_def - specify matching otherwise-unmatched multicast
+ * @spec: Specification to initialise
+ */
+static inline int efx_filter_set_mc_def(struct efx_filter_spec *spec)
+{
+ spec->match_flags |= EFX_FILTER_MATCH_LOC_MAC_IG;
+ spec->loc_mac[0] = 1;
+ return 0;
+}
+
#endif /* EFX_FILTER_H */
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [PATCH net-next 06/16] sfc: Name the RX drop queue ID
From: Ben Hutchings @ 2013-08-27 20:43 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377635844.13272.69.camel@bwh-desktop.uk.level5networks.com>
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/ethtool.c | 4 ++--
drivers/net/ethernet/sfc/filter.h | 7 ++++++-
2 files changed, 8 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/sfc/ethtool.c b/drivers/net/ethernet/sfc/ethtool.c
index 4db37f7..ec5cacd 100644
--- a/drivers/net/ethernet/sfc/ethtool.c
+++ b/drivers/net/ethernet/sfc/ethtool.c
@@ -822,7 +822,7 @@ static int efx_ethtool_get_class_rule(struct efx_nic *efx,
if (rc)
return rc;
- if (spec.dmaq_id == 0xfff)
+ if (spec.dmaq_id == EFX_FILTER_RX_DMAQ_ID_DROP)
rule->ring_cookie = RX_CLS_FLOW_DISC;
else
rule->ring_cookie = spec.dmaq_id;
@@ -967,7 +967,7 @@ static int efx_ethtool_set_class_rule(struct efx_nic *efx,
efx_filter_init_rx(&spec, EFX_FILTER_PRI_MANUAL,
efx->rx_scatter ? EFX_FILTER_FLAG_RX_SCATTER : 0,
(rule->ring_cookie == RX_CLS_FLOW_DISC) ?
- 0xfff : rule->ring_cookie);
+ EFX_FILTER_RX_DMAQ_ID_DROP : rule->ring_cookie);
switch (rule->flow_type) {
case TCP_V4_FLOW:
diff --git a/drivers/net/ethernet/sfc/filter.h b/drivers/net/ethernet/sfc/filter.h
index 5cb5472..b1170d4 100644
--- a/drivers/net/ethernet/sfc/filter.h
+++ b/drivers/net/ethernet/sfc/filter.h
@@ -76,7 +76,8 @@ enum efx_filter_flags {
* @type: Type of match to be performed, from &enum efx_filter_type
* @priority: Priority of the filter, from &enum efx_filter_priority
* @flags: Miscellaneous flags, from &enum efx_filter_flags
- * @dmaq_id: Source/target queue index
+ * @dmaq_id: Source/target queue index, or %EFX_FILTER_RX_DMAQ_ID_DROP for
+ * an RX drop filter
* @data: Match data (type-dependent)
*
* Use the efx_filter_set_*() functions to initialise the @type and
@@ -94,6 +95,10 @@ struct efx_filter_spec {
u32 data[3];
};
+enum {
+ EFX_FILTER_RX_DMAQ_ID_DROP = 0xfff
+};
+
static inline void efx_filter_init_rx(struct efx_filter_spec *spec,
enum efx_filter_priority priority,
enum efx_filter_flags flags,
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [PATCH net-next 05/16] sfc: Rename Falcon-arch filter implementation types and functions
From: Ben Hutchings @ 2013-08-27 20:43 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377635844.13272.69.camel@bwh-desktop.uk.level5networks.com>
The filter table(s) on EF10 are managed by firmware and will need
almost entirely separate code. Rename the types and functions used
within the existing implementation. The current definition of struct
efx_filter_spec is really implementation-specific, so we need to keep
it. For now, define a separate structure for the internal
representation but leave them identical.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/filter.c | 452 ++++++++++++++++++++------------------
1 file changed, 243 insertions(+), 209 deletions(-)
diff --git a/drivers/net/ethernet/sfc/filter.c b/drivers/net/ethernet/sfc/filter.c
index a588c95..c547630 100644
--- a/drivers/net/ethernet/sfc/filter.c
+++ b/drivers/net/ethernet/sfc/filter.c
@@ -19,60 +19,69 @@
* Due to pipelined implementation we need to program H/W with a value that
* is larger than the hop limit we want.
*/
-#define FILTER_CTL_SRCH_FUDGE_WILD 3
-#define FILTER_CTL_SRCH_FUDGE_FULL 1
+#define EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD 3
+#define EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL 1
/* Hard maximum hop limit. Hardware will time-out beyond 200-something.
- * We also need to avoid infinite loops in efx_filter_search() when the
+ * We also need to avoid infinite loops in efx_farch_filter_search() when the
* table is full.
*/
-#define FILTER_CTL_SRCH_MAX 200
+#define EFX_FARCH_FILTER_CTL_SRCH_MAX 200
/* Don't try very hard to find space for performance hints, as this is
* counter-productive. */
-#define FILTER_CTL_SRCH_HINT_MAX 5
-
-enum efx_filter_table_id {
- EFX_FILTER_TABLE_RX_IP = 0,
- EFX_FILTER_TABLE_RX_MAC,
- EFX_FILTER_TABLE_RX_DEF,
- EFX_FILTER_TABLE_TX_MAC,
- EFX_FILTER_TABLE_COUNT,
+#define EFX_FARCH_FILTER_CTL_SRCH_HINT_MAX 5
+
+enum efx_farch_filter_table_id {
+ EFX_FARCH_FILTER_TABLE_RX_IP = 0,
+ EFX_FARCH_FILTER_TABLE_RX_MAC,
+ EFX_FARCH_FILTER_TABLE_RX_DEF,
+ EFX_FARCH_FILTER_TABLE_TX_MAC,
+ EFX_FARCH_FILTER_TABLE_COUNT,
};
-enum efx_filter_index {
- EFX_FILTER_INDEX_UC_DEF,
- EFX_FILTER_INDEX_MC_DEF,
- EFX_FILTER_SIZE_RX_DEF,
+enum efx_farch_filter_index {
+ EFX_FARCH_FILTER_INDEX_UC_DEF,
+ EFX_FARCH_FILTER_INDEX_MC_DEF,
+ EFX_FARCH_FILTER_SIZE_RX_DEF,
};
-struct efx_filter_table {
- enum efx_filter_table_id id;
+struct efx_farch_filter_spec {
+ u8 type:4;
+ u8 priority:4;
+ u8 flags;
+ u16 dmaq_id;
+ u32 data[3];
+};
+
+struct efx_farch_filter_table {
+ enum efx_farch_filter_table_id id;
u32 offset; /* address of table relative to BAR */
unsigned size; /* number of entries */
unsigned step; /* step between entries */
unsigned used; /* number currently used */
unsigned long *used_bitmap;
- struct efx_filter_spec *spec;
+ struct efx_farch_filter_spec *spec;
unsigned search_depth[EFX_FILTER_TYPE_COUNT];
};
struct efx_filter_state {
spinlock_t lock;
- struct efx_filter_table table[EFX_FILTER_TABLE_COUNT];
+ struct efx_farch_filter_table table[EFX_FARCH_FILTER_TABLE_COUNT];
#ifdef CONFIG_RFS_ACCEL
u32 *rps_flow_id;
unsigned rps_expire_index;
#endif
};
-static void efx_filter_table_clear_entry(struct efx_nic *efx,
- struct efx_filter_table *table,
- unsigned int filter_idx);
+static void
+efx_farch_filter_table_clear_entry(struct efx_nic *efx,
+ struct efx_farch_filter_table *table,
+ unsigned int filter_idx);
/* The filter hash function is LFSR polynomial x^16 + x^3 + 1 of a 32-bit
* key derived from the n-tuple. The initial LFSR state is 0xffff. */
-static u16 efx_filter_hash(u32 key)
+static u16 efx_farch_filter_hash(u32 key)
{
u16 tmp;
@@ -88,89 +97,97 @@ static u16 efx_filter_hash(u32 key)
/* To allow for hash collisions, filter search continues at these
* increments from the first possible entry selected by the hash. */
-static u16 efx_filter_increment(u32 key)
+static u16 efx_farch_filter_increment(u32 key)
{
return key * 2 - 1;
}
-static enum efx_filter_table_id
-efx_filter_spec_table_id(const struct efx_filter_spec *spec)
+static enum efx_farch_filter_table_id
+efx_farch_filter_spec_table_id(const struct efx_farch_filter_spec *spec)
{
- BUILD_BUG_ON(EFX_FILTER_TABLE_RX_IP != (EFX_FILTER_TCP_FULL >> 2));
- BUILD_BUG_ON(EFX_FILTER_TABLE_RX_IP != (EFX_FILTER_TCP_WILD >> 2));
- BUILD_BUG_ON(EFX_FILTER_TABLE_RX_IP != (EFX_FILTER_UDP_FULL >> 2));
- BUILD_BUG_ON(EFX_FILTER_TABLE_RX_IP != (EFX_FILTER_UDP_WILD >> 2));
- BUILD_BUG_ON(EFX_FILTER_TABLE_RX_MAC != (EFX_FILTER_MAC_FULL >> 2));
- BUILD_BUG_ON(EFX_FILTER_TABLE_RX_MAC != (EFX_FILTER_MAC_WILD >> 2));
- BUILD_BUG_ON(EFX_FILTER_TABLE_TX_MAC != EFX_FILTER_TABLE_RX_MAC + 2);
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
+ (EFX_FILTER_TCP_FULL >> 2));
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
+ (EFX_FILTER_TCP_WILD >> 2));
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
+ (EFX_FILTER_UDP_FULL >> 2));
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_IP !=
+ (EFX_FILTER_UDP_WILD >> 2));
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_MAC !=
+ (EFX_FILTER_MAC_FULL >> 2));
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_RX_MAC !=
+ (EFX_FILTER_MAC_WILD >> 2));
+ BUILD_BUG_ON(EFX_FARCH_FILTER_TABLE_TX_MAC !=
+ EFX_FARCH_FILTER_TABLE_RX_MAC + 2);
EFX_BUG_ON_PARANOID(spec->type == EFX_FILTER_UNSPEC);
return (spec->type >> 2) + ((spec->flags & EFX_FILTER_FLAG_TX) ? 2 : 0);
}
-static struct efx_filter_table *
-efx_filter_spec_table(struct efx_filter_state *state,
- const struct efx_filter_spec *spec)
+static struct efx_farch_filter_table *
+efx_farch_filter_spec_table(struct efx_filter_state *state,
+ const struct efx_farch_filter_spec *spec)
{
if (spec->type == EFX_FILTER_UNSPEC)
return NULL;
else
- return &state->table[efx_filter_spec_table_id(spec)];
+ return &state->table[efx_farch_filter_spec_table_id(spec)];
}
-static void efx_filter_table_reset_search_depth(struct efx_filter_table *table)
+static void
+efx_farch_filter_table_reset_search_depth(struct efx_farch_filter_table *table)
{
memset(table->search_depth, 0, sizeof(table->search_depth));
}
-static void efx_filter_push_rx_config(struct efx_nic *efx)
+static void efx_farch_filter_push_rx_config(struct efx_nic *efx)
{
struct efx_filter_state *state = efx->filter_state;
- struct efx_filter_table *table;
+ struct efx_farch_filter_table *table;
efx_oword_t filter_ctl;
efx_reado(efx, &filter_ctl, FR_BZ_RX_FILTER_CTL);
- table = &state->table[EFX_FILTER_TABLE_RX_IP];
+ table = &state->table[EFX_FARCH_FILTER_TABLE_RX_IP];
EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_TCP_FULL_SRCH_LIMIT,
table->search_depth[EFX_FILTER_TCP_FULL] +
- FILTER_CTL_SRCH_FUDGE_FULL);
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_TCP_WILD_SRCH_LIMIT,
table->search_depth[EFX_FILTER_TCP_WILD] +
- FILTER_CTL_SRCH_FUDGE_WILD);
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_UDP_FULL_SRCH_LIMIT,
table->search_depth[EFX_FILTER_UDP_FULL] +
- FILTER_CTL_SRCH_FUDGE_FULL);
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
EFX_SET_OWORD_FIELD(filter_ctl, FRF_BZ_UDP_WILD_SRCH_LIMIT,
table->search_depth[EFX_FILTER_UDP_WILD] +
- FILTER_CTL_SRCH_FUDGE_WILD);
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
- table = &state->table[EFX_FILTER_TABLE_RX_MAC];
+ table = &state->table[EFX_FARCH_FILTER_TABLE_RX_MAC];
if (table->size) {
EFX_SET_OWORD_FIELD(
filter_ctl, FRF_CZ_ETHERNET_FULL_SEARCH_LIMIT,
table->search_depth[EFX_FILTER_MAC_FULL] +
- FILTER_CTL_SRCH_FUDGE_FULL);
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
EFX_SET_OWORD_FIELD(
filter_ctl, FRF_CZ_ETHERNET_WILDCARD_SEARCH_LIMIT,
table->search_depth[EFX_FILTER_MAC_WILD] +
- FILTER_CTL_SRCH_FUDGE_WILD);
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
}
- table = &state->table[EFX_FILTER_TABLE_RX_DEF];
+ table = &state->table[EFX_FARCH_FILTER_TABLE_RX_DEF];
if (table->size) {
EFX_SET_OWORD_FIELD(
filter_ctl, FRF_CZ_UNICAST_NOMATCH_Q_ID,
- table->spec[EFX_FILTER_INDEX_UC_DEF].dmaq_id);
+ table->spec[EFX_FARCH_FILTER_INDEX_UC_DEF].dmaq_id);
EFX_SET_OWORD_FIELD(
filter_ctl, FRF_CZ_UNICAST_NOMATCH_RSS_ENABLED,
- !!(table->spec[EFX_FILTER_INDEX_UC_DEF].flags &
+ !!(table->spec[EFX_FARCH_FILTER_INDEX_UC_DEF].flags &
EFX_FILTER_FLAG_RX_RSS));
EFX_SET_OWORD_FIELD(
filter_ctl, FRF_CZ_MULTICAST_NOMATCH_Q_ID,
- table->spec[EFX_FILTER_INDEX_MC_DEF].dmaq_id);
+ table->spec[EFX_FARCH_FILTER_INDEX_MC_DEF].dmaq_id);
EFX_SET_OWORD_FIELD(
filter_ctl, FRF_CZ_MULTICAST_NOMATCH_RSS_ENABLED,
- !!(table->spec[EFX_FILTER_INDEX_MC_DEF].flags &
+ !!(table->spec[EFX_FARCH_FILTER_INDEX_MC_DEF].flags &
EFX_FILTER_FLAG_RX_RSS));
/* There is a single bit to enable RX scatter for all
@@ -179,8 +196,8 @@ static void efx_filter_push_rx_config(struct efx_nic *efx)
*/
EFX_SET_OWORD_FIELD(
filter_ctl, FRF_BZ_SCATTER_ENBL_NO_MATCH_Q,
- !!(table->spec[EFX_FILTER_INDEX_UC_DEF].flags &
- table->spec[EFX_FILTER_INDEX_MC_DEF].flags &
+ !!(table->spec[EFX_FARCH_FILTER_INDEX_UC_DEF].flags &
+ table->spec[EFX_FARCH_FILTER_INDEX_MC_DEF].flags &
EFX_FILTER_FLAG_RX_SCATTER));
} else if (efx_nic_rev(efx) >= EFX_REV_FALCON_B0) {
/* We don't expose 'default' filters because unmatched
@@ -196,24 +213,24 @@ static void efx_filter_push_rx_config(struct efx_nic *efx)
efx_writeo(efx, &filter_ctl, FR_BZ_RX_FILTER_CTL);
}
-static void efx_filter_push_tx_limits(struct efx_nic *efx)
+static void efx_farch_filter_push_tx_limits(struct efx_nic *efx)
{
struct efx_filter_state *state = efx->filter_state;
- struct efx_filter_table *table;
+ struct efx_farch_filter_table *table;
efx_oword_t tx_cfg;
efx_reado(efx, &tx_cfg, FR_AZ_TX_CFG);
- table = &state->table[EFX_FILTER_TABLE_TX_MAC];
+ table = &state->table[EFX_FARCH_FILTER_TABLE_TX_MAC];
if (table->size) {
EFX_SET_OWORD_FIELD(
tx_cfg, FRF_CZ_TX_ETH_FILTER_FULL_SEARCH_RANGE,
table->search_depth[EFX_FILTER_MAC_FULL] +
- FILTER_CTL_SRCH_FUDGE_FULL);
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_FULL);
EFX_SET_OWORD_FIELD(
tx_cfg, FRF_CZ_TX_ETH_FILTER_WILD_SEARCH_RANGE,
table->search_depth[EFX_FILTER_MAC_WILD] +
- FILTER_CTL_SRCH_FUDGE_WILD);
+ EFX_FARCH_FILTER_CTL_SRCH_FUDGE_WILD);
}
efx_writeo(efx, &tx_cfg, FR_AZ_TX_CFG);
@@ -427,24 +444,23 @@ int efx_filter_set_mc_def(struct efx_filter_spec *spec)
return 0;
}
-static void efx_filter_reset_rx_def(struct efx_nic *efx, unsigned filter_idx)
+static void
+efx_farch_filter_reset_rx_def(struct efx_nic *efx, unsigned filter_idx)
{
struct efx_filter_state *state = efx->filter_state;
- struct efx_filter_table *table = &state->table[EFX_FILTER_TABLE_RX_DEF];
- struct efx_filter_spec *spec = &table->spec[filter_idx];
- enum efx_filter_flags flags = 0;
+ struct efx_farch_filter_table *table =
+ &state->table[EFX_FARCH_FILTER_TABLE_RX_DEF];
+ struct efx_farch_filter_spec *spec = &table->spec[filter_idx];
/* If there's only one channel then disable RSS for non VF
* traffic, thereby allowing VFs to use RSS when the PF can't.
*/
- if (efx->n_rx_channels > 1)
- flags |= EFX_FILTER_FLAG_RX_RSS;
-
- if (efx->rx_scatter)
- flags |= EFX_FILTER_FLAG_RX_SCATTER;
-
- efx_filter_init_rx(spec, EFX_FILTER_PRI_MANUAL, flags, 0);
spec->type = EFX_FILTER_UC_DEF + filter_idx;
+ spec->priority = EFX_FILTER_PRI_MANUAL;
+ spec->flags = (EFX_FILTER_FLAG_RX |
+ (efx->n_rx_channels > 1 ? EFX_FILTER_FLAG_RX_RSS : 0) |
+ (efx->rx_scatter ? EFX_FILTER_FLAG_RX_SCATTER : 0));
+ spec->dmaq_id = 0;
table->used_bitmap[0] |= 1 << filter_idx;
}
@@ -472,12 +488,13 @@ int efx_filter_get_eth_local(const struct efx_filter_spec *spec,
}
/* Build a filter entry and return its n-tuple key. */
-static u32 efx_filter_build(efx_oword_t *filter, struct efx_filter_spec *spec)
+static u32 efx_farch_filter_build(efx_oword_t *filter,
+ struct efx_farch_filter_spec *spec)
{
u32 data3;
- switch (efx_filter_spec_table_id(spec)) {
- case EFX_FILTER_TABLE_RX_IP: {
+ switch (efx_farch_filter_spec_table_id(spec)) {
+ case EFX_FARCH_FILTER_TABLE_RX_IP: {
bool is_udp = (spec->type == EFX_FILTER_UDP_FULL ||
spec->type == EFX_FILTER_UDP_WILD);
EFX_POPULATE_OWORD_7(
@@ -495,7 +512,7 @@ static u32 efx_filter_build(efx_oword_t *filter, struct efx_filter_spec *spec)
break;
}
- case EFX_FILTER_TABLE_RX_MAC: {
+ case EFX_FARCH_FILTER_TABLE_RX_MAC: {
bool is_wild = spec->type == EFX_FILTER_MAC_WILD;
EFX_POPULATE_OWORD_7(
*filter,
@@ -512,7 +529,7 @@ static u32 efx_filter_build(efx_oword_t *filter, struct efx_filter_spec *spec)
break;
}
- case EFX_FILTER_TABLE_TX_MAC: {
+ case EFX_FARCH_FILTER_TABLE_TX_MAC: {
bool is_wild = spec->type == EFX_FILTER_MAC_WILD;
EFX_POPULATE_OWORD_5(*filter,
FRF_CZ_TMFT_TXQ_ID, spec->dmaq_id,
@@ -531,8 +548,8 @@ static u32 efx_filter_build(efx_oword_t *filter, struct efx_filter_spec *spec)
return spec->data[0] ^ spec->data[1] ^ spec->data[2] ^ data3;
}
-static bool efx_filter_equal(const struct efx_filter_spec *left,
- const struct efx_filter_spec *right)
+static bool efx_farch_filter_equal(const struct efx_farch_filter_spec *left,
+ const struct efx_farch_filter_spec *right)
{
if (left->type != right->type ||
memcmp(left->data, right->data, sizeof(left->data)))
@@ -554,9 +571,9 @@ static bool efx_filter_equal(const struct efx_filter_spec *left,
* accept user-provided IDs.
*/
-#define EFX_FILTER_MATCH_PRI_COUNT 5
+#define EFX_FARCH_FILTER_MATCH_PRI_COUNT 5
-static const u8 efx_filter_type_match_pri[EFX_FILTER_TYPE_COUNT] = {
+static const u8 efx_farch_filter_type_match_pri[EFX_FILTER_TYPE_COUNT] = {
[EFX_FILTER_TCP_FULL] = 0,
[EFX_FILTER_UDP_FULL] = 0,
[EFX_FILTER_TCP_WILD] = 1,
@@ -567,58 +584,58 @@ static const u8 efx_filter_type_match_pri[EFX_FILTER_TYPE_COUNT] = {
[EFX_FILTER_MC_DEF] = 4,
};
-static const enum efx_filter_table_id efx_filter_range_table[] = {
- EFX_FILTER_TABLE_RX_IP, /* RX match pri 0 */
- EFX_FILTER_TABLE_RX_IP,
- EFX_FILTER_TABLE_RX_MAC,
- EFX_FILTER_TABLE_RX_MAC,
- EFX_FILTER_TABLE_RX_DEF, /* RX match pri 4 */
- EFX_FILTER_TABLE_COUNT, /* TX match pri 0; invalid */
- EFX_FILTER_TABLE_COUNT, /* invalid */
- EFX_FILTER_TABLE_TX_MAC,
- EFX_FILTER_TABLE_TX_MAC, /* TX match pri 3 */
+static const enum efx_farch_filter_table_id efx_farch_filter_range_table[] = {
+ EFX_FARCH_FILTER_TABLE_RX_IP, /* RX match pri 0 */
+ EFX_FARCH_FILTER_TABLE_RX_IP,
+ EFX_FARCH_FILTER_TABLE_RX_MAC,
+ EFX_FARCH_FILTER_TABLE_RX_MAC,
+ EFX_FARCH_FILTER_TABLE_RX_DEF, /* RX match pri 4 */
+ EFX_FARCH_FILTER_TABLE_TX_MAC, /* TX match pri 0 */
+ EFX_FARCH_FILTER_TABLE_TX_MAC, /* TX match pri 1 */
};
-#define EFX_FILTER_INDEX_WIDTH 13
-#define EFX_FILTER_INDEX_MASK ((1 << EFX_FILTER_INDEX_WIDTH) - 1)
+#define EFX_FARCH_FILTER_INDEX_WIDTH 13
+#define EFX_FARCH_FILTER_INDEX_MASK ((1 << EFX_FARCH_FILTER_INDEX_WIDTH) - 1)
static inline u32
-efx_filter_make_id(const struct efx_filter_spec *spec, unsigned int index)
+efx_farch_filter_make_id(const struct efx_farch_filter_spec *spec,
+ unsigned int index)
{
unsigned int range;
- range = efx_filter_type_match_pri[spec->type];
+ range = efx_farch_filter_type_match_pri[spec->type];
if (!(spec->flags & EFX_FILTER_FLAG_RX))
- range += EFX_FILTER_MATCH_PRI_COUNT;
+ range += EFX_FARCH_FILTER_MATCH_PRI_COUNT;
- return range << EFX_FILTER_INDEX_WIDTH | index;
+ return range << EFX_FARCH_FILTER_INDEX_WIDTH | index;
}
-static inline enum efx_filter_table_id efx_filter_id_table_id(u32 id)
+static inline enum efx_farch_filter_table_id
+efx_farch_filter_id_table_id(u32 id)
{
- unsigned int range = id >> EFX_FILTER_INDEX_WIDTH;
+ unsigned int range = id >> EFX_FARCH_FILTER_INDEX_WIDTH;
- if (range < ARRAY_SIZE(efx_filter_range_table))
- return efx_filter_range_table[range];
+ if (range < ARRAY_SIZE(efx_farch_filter_range_table))
+ return efx_farch_filter_range_table[range];
else
- return EFX_FILTER_TABLE_COUNT; /* invalid */
+ return EFX_FARCH_FILTER_TABLE_COUNT; /* invalid */
}
-static inline unsigned int efx_filter_id_index(u32 id)
+static inline unsigned int efx_farch_filter_id_index(u32 id)
{
- return id & EFX_FILTER_INDEX_MASK;
+ return id & EFX_FARCH_FILTER_INDEX_MASK;
}
u32 efx_filter_get_rx_id_limit(struct efx_nic *efx)
{
struct efx_filter_state *state = efx->filter_state;
- unsigned int range = EFX_FILTER_MATCH_PRI_COUNT - 1;
- enum efx_filter_table_id table_id;
+ unsigned int range = EFX_FARCH_FILTER_MATCH_PRI_COUNT - 1;
+ enum efx_farch_filter_table_id table_id;
do {
- table_id = efx_filter_range_table[range];
+ table_id = efx_farch_filter_range_table[range];
if (state->table[table_id].size != 0)
- return range << EFX_FILTER_INDEX_WIDTH |
+ return range << EFX_FARCH_FILTER_INDEX_WIDTH |
state->table[table_id].size;
} while (range--);
@@ -643,29 +660,35 @@ u32 efx_filter_get_rx_id_limit(struct efx_nic *efx)
* the existing filter has higher priority or -%EEXIST if it has
* equal priority.
*/
-s32 efx_filter_insert_filter(struct efx_nic *efx, struct efx_filter_spec *spec,
+s32 efx_filter_insert_filter(struct efx_nic *efx,
+ struct efx_filter_spec *gen_spec,
bool replace_equal)
{
struct efx_filter_state *state = efx->filter_state;
- struct efx_filter_table *table = efx_filter_spec_table(state, spec);
+ struct efx_farch_filter_table *table;
+ struct efx_farch_filter_spec spec;
efx_oword_t filter;
int rep_index, ins_index;
unsigned int depth = 0;
int rc;
+ /* XXX efx_farch_filter_spec and efx_filter_spec will diverge in future */
+ memcpy(&spec, gen_spec, sizeof(*gen_spec));
+
+ table = efx_farch_filter_spec_table(state, &spec);
if (!table || table->size == 0)
return -EINVAL;
netif_vdbg(efx, hw, efx->net_dev,
- "%s: type %d search_depth=%d", __func__, spec->type,
- table->search_depth[spec->type]);
+ "%s: type %d search_depth=%d", __func__, spec.type,
+ table->search_depth[spec.type]);
- if (table->id == EFX_FILTER_TABLE_RX_DEF) {
+ if (table->id == EFX_FARCH_FILTER_TABLE_RX_DEF) {
/* One filter spec per type */
- BUILD_BUG_ON(EFX_FILTER_INDEX_UC_DEF != 0);
- BUILD_BUG_ON(EFX_FILTER_INDEX_MC_DEF !=
+ BUILD_BUG_ON(EFX_FARCH_FILTER_INDEX_UC_DEF != 0);
+ BUILD_BUG_ON(EFX_FARCH_FILTER_INDEX_MC_DEF !=
EFX_FILTER_MC_DEF - EFX_FILTER_UC_DEF);
- rep_index = spec->type - EFX_FILTER_UC_DEF;
+ rep_index = spec.type - EFX_FILTER_UC_DEF;
ins_index = rep_index;
spin_lock_bh(&state->lock);
@@ -685,13 +708,14 @@ s32 efx_filter_insert_filter(struct efx_nic *efx, struct efx_filter_spec *spec,
* (b) we have searched exhaustively for (1), and have
* either found (2) or searched exhaustively for it
*/
- u32 key = efx_filter_build(&filter, spec);
- unsigned int hash = efx_filter_hash(key);
- unsigned int incr = efx_filter_increment(key);
- unsigned int max_rep_depth = table->search_depth[spec->type];
+ u32 key = efx_farch_filter_build(&filter, &spec);
+ unsigned int hash = efx_farch_filter_hash(key);
+ unsigned int incr = efx_farch_filter_increment(key);
+ unsigned int max_rep_depth = table->search_depth[spec.type];
unsigned int max_ins_depth =
- spec->priority <= EFX_FILTER_PRI_HINT ?
- FILTER_CTL_SRCH_HINT_MAX : FILTER_CTL_SRCH_MAX;
+ spec.priority <= EFX_FILTER_PRI_HINT ?
+ EFX_FARCH_FILTER_CTL_SRCH_HINT_MAX :
+ EFX_FARCH_FILTER_CTL_SRCH_MAX;
unsigned int i = hash & (table->size - 1);
ins_index = -1;
@@ -703,7 +727,8 @@ s32 efx_filter_insert_filter(struct efx_nic *efx, struct efx_filter_spec *spec,
if (!test_bit(i, table->used_bitmap)) {
if (ins_index < 0)
ins_index = i;
- } else if (efx_filter_equal(spec, &table->spec[i])) {
+ } else if (efx_farch_filter_equal(&spec,
+ &table->spec[i])) {
/* Case (a) */
if (ins_index < 0)
ins_index = i;
@@ -731,13 +756,14 @@ s32 efx_filter_insert_filter(struct efx_nic *efx, struct efx_filter_spec *spec,
* should do so
*/
if (rep_index >= 0) {
- struct efx_filter_spec *saved_spec = &table->spec[rep_index];
+ struct efx_farch_filter_spec *saved_spec =
+ &table->spec[rep_index];
- if (spec->priority == saved_spec->priority && !replace_equal) {
+ if (spec.priority == saved_spec->priority && !replace_equal) {
rc = -EEXIST;
goto out;
}
- if (spec->priority < saved_spec->priority) {
+ if (spec.priority < saved_spec->priority) {
rc = -EPERM;
goto out;
}
@@ -748,17 +774,17 @@ s32 efx_filter_insert_filter(struct efx_nic *efx, struct efx_filter_spec *spec,
__set_bit(ins_index, table->used_bitmap);
++table->used;
}
- table->spec[ins_index] = *spec;
+ table->spec[ins_index] = spec;
- if (table->id == EFX_FILTER_TABLE_RX_DEF) {
- efx_filter_push_rx_config(efx);
+ if (table->id == EFX_FARCH_FILTER_TABLE_RX_DEF) {
+ efx_farch_filter_push_rx_config(efx);
} else {
- if (table->search_depth[spec->type] < depth) {
- table->search_depth[spec->type] = depth;
- if (spec->flags & EFX_FILTER_FLAG_TX)
- efx_filter_push_tx_limits(efx);
+ if (table->search_depth[spec.type] < depth) {
+ table->search_depth[spec.type] = depth;
+ if (spec.flags & EFX_FILTER_FLAG_TX)
+ efx_farch_filter_push_tx_limits(efx);
else
- efx_filter_push_rx_config(efx);
+ efx_farch_filter_push_rx_config(efx);
}
efx_writeo(efx, &filter,
@@ -768,29 +794,31 @@ s32 efx_filter_insert_filter(struct efx_nic *efx, struct efx_filter_spec *spec,
* at a lower depth, clear the replaced filter
*/
if (ins_index != rep_index && rep_index >= 0)
- efx_filter_table_clear_entry(efx, table, rep_index);
+ efx_farch_filter_table_clear_entry(efx, table,
+ rep_index);
}
netif_vdbg(efx, hw, efx->net_dev,
"%s: filter type %d index %d rxq %u set",
- __func__, spec->type, ins_index, spec->dmaq_id);
- rc = efx_filter_make_id(spec, ins_index);
+ __func__, spec.type, ins_index, spec.dmaq_id);
+ rc = efx_farch_filter_make_id(&spec, ins_index);
out:
spin_unlock_bh(&state->lock);
return rc;
}
-static void efx_filter_table_clear_entry(struct efx_nic *efx,
- struct efx_filter_table *table,
- unsigned int filter_idx)
+static void
+efx_farch_filter_table_clear_entry(struct efx_nic *efx,
+ struct efx_farch_filter_table *table,
+ unsigned int filter_idx)
{
static efx_oword_t filter;
- if (table->id == EFX_FILTER_TABLE_RX_DEF) {
+ if (table->id == EFX_FARCH_FILTER_TABLE_RX_DEF) {
/* RX default filters must always exist */
- efx_filter_reset_rx_def(efx, filter_idx);
- efx_filter_push_rx_config(efx);
+ efx_farch_filter_reset_rx_def(efx, filter_idx);
+ efx_farch_filter_push_rx_config(efx);
} else if (test_bit(filter_idx, table->used_bitmap)) {
__clear_bit(filter_idx, table->used_bitmap);
--table->used;
@@ -815,18 +843,18 @@ int efx_filter_remove_id_safe(struct efx_nic *efx,
u32 filter_id)
{
struct efx_filter_state *state = efx->filter_state;
- enum efx_filter_table_id table_id;
- struct efx_filter_table *table;
+ enum efx_farch_filter_table_id table_id;
+ struct efx_farch_filter_table *table;
unsigned int filter_idx;
- struct efx_filter_spec *spec;
+ struct efx_farch_filter_spec *spec;
int rc;
- table_id = efx_filter_id_table_id(filter_id);
- if ((unsigned int)table_id >= EFX_FILTER_TABLE_COUNT)
+ table_id = efx_farch_filter_id_table_id(filter_id);
+ if ((unsigned int)table_id >= EFX_FARCH_FILTER_TABLE_COUNT)
return -ENOENT;
table = &state->table[table_id];
- filter_idx = efx_filter_id_index(filter_id);
+ filter_idx = efx_farch_filter_id_index(filter_id);
if (filter_idx >= table->size)
return -ENOENT;
spec = &table->spec[filter_idx];
@@ -835,9 +863,9 @@ int efx_filter_remove_id_safe(struct efx_nic *efx,
if (test_bit(filter_idx, table->used_bitmap) &&
spec->priority == priority) {
- efx_filter_table_clear_entry(efx, table, filter_idx);
+ efx_farch_filter_table_clear_entry(efx, table, filter_idx);
if (table->used == 0)
- efx_filter_table_reset_search_depth(table);
+ efx_farch_filter_table_reset_search_depth(table);
rc = 0;
} else {
rc = -ENOENT;
@@ -863,18 +891,18 @@ int efx_filter_get_filter_safe(struct efx_nic *efx,
u32 filter_id, struct efx_filter_spec *spec_buf)
{
struct efx_filter_state *state = efx->filter_state;
- enum efx_filter_table_id table_id;
- struct efx_filter_table *table;
- struct efx_filter_spec *spec;
+ enum efx_farch_filter_table_id table_id;
+ struct efx_farch_filter_table *table;
+ struct efx_farch_filter_spec *spec;
unsigned int filter_idx;
int rc;
- table_id = efx_filter_id_table_id(filter_id);
- if ((unsigned int)table_id >= EFX_FILTER_TABLE_COUNT)
+ table_id = efx_farch_filter_id_table_id(filter_id);
+ if ((unsigned int)table_id >= EFX_FARCH_FILTER_TABLE_COUNT)
return -ENOENT;
table = &state->table[table_id];
- filter_idx = efx_filter_id_index(filter_id);
+ filter_idx = efx_farch_filter_id_index(filter_id);
if (filter_idx >= table->size)
return -ENOENT;
spec = &table->spec[filter_idx];
@@ -883,7 +911,8 @@ int efx_filter_get_filter_safe(struct efx_nic *efx,
if (test_bit(filter_idx, table->used_bitmap) &&
spec->priority == priority) {
- *spec_buf = *spec;
+ /* XXX efx_farch_filter_spec and efx_filter_spec will diverge */
+ memcpy(spec_buf, spec, sizeof(*spec));
rc = 0;
} else {
rc = -ENOENT;
@@ -894,21 +923,23 @@ int efx_filter_get_filter_safe(struct efx_nic *efx,
return rc;
}
-static void efx_filter_table_clear(struct efx_nic *efx,
- enum efx_filter_table_id table_id,
- enum efx_filter_priority priority)
+static void
+efx_farch_filter_table_clear(struct efx_nic *efx,
+ enum efx_farch_filter_table_id table_id,
+ enum efx_filter_priority priority)
{
struct efx_filter_state *state = efx->filter_state;
- struct efx_filter_table *table = &state->table[table_id];
+ struct efx_farch_filter_table *table = &state->table[table_id];
unsigned int filter_idx;
spin_lock_bh(&state->lock);
for (filter_idx = 0; filter_idx < table->size; ++filter_idx)
if (table->spec[filter_idx].priority <= priority)
- efx_filter_table_clear_entry(efx, table, filter_idx);
+ efx_farch_filter_table_clear_entry(efx, table,
+ filter_idx);
if (table->used == 0)
- efx_filter_table_reset_search_depth(table);
+ efx_farch_filter_table_reset_search_depth(table);
spin_unlock_bh(&state->lock);
}
@@ -920,23 +951,25 @@ static void efx_filter_table_clear(struct efx_nic *efx,
*/
void efx_filter_clear_rx(struct efx_nic *efx, enum efx_filter_priority priority)
{
- efx_filter_table_clear(efx, EFX_FILTER_TABLE_RX_IP, priority);
- efx_filter_table_clear(efx, EFX_FILTER_TABLE_RX_MAC, priority);
+ efx_farch_filter_table_clear(efx, EFX_FARCH_FILTER_TABLE_RX_IP,
+ priority);
+ efx_farch_filter_table_clear(efx, EFX_FARCH_FILTER_TABLE_RX_MAC,
+ priority);
}
u32 efx_filter_count_rx_used(struct efx_nic *efx,
enum efx_filter_priority priority)
{
struct efx_filter_state *state = efx->filter_state;
- enum efx_filter_table_id table_id;
- struct efx_filter_table *table;
+ enum efx_farch_filter_table_id table_id;
+ struct efx_farch_filter_table *table;
unsigned int filter_idx;
u32 count = 0;
spin_lock_bh(&state->lock);
- for (table_id = EFX_FILTER_TABLE_RX_IP;
- table_id <= EFX_FILTER_TABLE_RX_DEF;
+ for (table_id = EFX_FARCH_FILTER_TABLE_RX_IP;
+ table_id <= EFX_FARCH_FILTER_TABLE_RX_DEF;
table_id++) {
table = &state->table[table_id];
for (filter_idx = 0; filter_idx < table->size; filter_idx++) {
@@ -956,15 +989,15 @@ s32 efx_filter_get_rx_ids(struct efx_nic *efx,
u32 *buf, u32 size)
{
struct efx_filter_state *state = efx->filter_state;
- enum efx_filter_table_id table_id;
- struct efx_filter_table *table;
+ enum efx_farch_filter_table_id table_id;
+ struct efx_farch_filter_table *table;
unsigned int filter_idx;
s32 count = 0;
spin_lock_bh(&state->lock);
- for (table_id = EFX_FILTER_TABLE_RX_IP;
- table_id <= EFX_FILTER_TABLE_RX_DEF;
+ for (table_id = EFX_FARCH_FILTER_TABLE_RX_IP;
+ table_id <= EFX_FARCH_FILTER_TABLE_RX_DEF;
table_id++) {
table = &state->table[table_id];
for (filter_idx = 0; filter_idx < table->size; filter_idx++) {
@@ -974,7 +1007,7 @@ s32 efx_filter_get_rx_ids(struct efx_nic *efx,
count = -EMSGSIZE;
goto out;
}
- buf[count++] = efx_filter_make_id(
+ buf[count++] = efx_farch_filter_make_id(
&table->spec[filter_idx], filter_idx);
}
}
@@ -989,14 +1022,14 @@ out:
void efx_restore_filters(struct efx_nic *efx)
{
struct efx_filter_state *state = efx->filter_state;
- enum efx_filter_table_id table_id;
- struct efx_filter_table *table;
+ enum efx_farch_filter_table_id table_id;
+ struct efx_farch_filter_table *table;
efx_oword_t filter;
unsigned int filter_idx;
spin_lock_bh(&state->lock);
- for (table_id = 0; table_id < EFX_FILTER_TABLE_COUNT; table_id++) {
+ for (table_id = 0; table_id < EFX_FARCH_FILTER_TABLE_COUNT; table_id++) {
table = &state->table[table_id];
/* Check whether this is a regular register table */
@@ -1006,14 +1039,14 @@ void efx_restore_filters(struct efx_nic *efx)
for (filter_idx = 0; filter_idx < table->size; filter_idx++) {
if (!test_bit(filter_idx, table->used_bitmap))
continue;
- efx_filter_build(&filter, &table->spec[filter_idx]);
+ efx_farch_filter_build(&filter, &table->spec[filter_idx]);
efx_writeo(efx, &filter,
table->offset + table->step * filter_idx);
}
}
- efx_filter_push_rx_config(efx);
- efx_filter_push_tx_limits(efx);
+ efx_farch_filter_push_rx_config(efx);
+ efx_farch_filter_push_tx_limits(efx);
spin_unlock_bh(&state->lock);
}
@@ -1021,7 +1054,7 @@ void efx_restore_filters(struct efx_nic *efx)
int efx_probe_filters(struct efx_nic *efx)
{
struct efx_filter_state *state;
- struct efx_filter_table *table;
+ struct efx_farch_filter_table *table;
unsigned table_id;
state = kzalloc(sizeof(*efx->filter_state), GFP_KERNEL);
@@ -1039,32 +1072,32 @@ int efx_probe_filters(struct efx_nic *efx)
if (!state->rps_flow_id)
goto fail;
#endif
- table = &state->table[EFX_FILTER_TABLE_RX_IP];
- table->id = EFX_FILTER_TABLE_RX_IP;
+ table = &state->table[EFX_FARCH_FILTER_TABLE_RX_IP];
+ table->id = EFX_FARCH_FILTER_TABLE_RX_IP;
table->offset = FR_BZ_RX_FILTER_TBL0;
table->size = FR_BZ_RX_FILTER_TBL0_ROWS;
table->step = FR_BZ_RX_FILTER_TBL0_STEP;
}
if (efx_nic_rev(efx) >= EFX_REV_SIENA_A0) {
- table = &state->table[EFX_FILTER_TABLE_RX_MAC];
- table->id = EFX_FILTER_TABLE_RX_MAC;
+ table = &state->table[EFX_FARCH_FILTER_TABLE_RX_MAC];
+ table->id = EFX_FARCH_FILTER_TABLE_RX_MAC;
table->offset = FR_CZ_RX_MAC_FILTER_TBL0;
table->size = FR_CZ_RX_MAC_FILTER_TBL0_ROWS;
table->step = FR_CZ_RX_MAC_FILTER_TBL0_STEP;
- table = &state->table[EFX_FILTER_TABLE_RX_DEF];
- table->id = EFX_FILTER_TABLE_RX_DEF;
- table->size = EFX_FILTER_SIZE_RX_DEF;
+ table = &state->table[EFX_FARCH_FILTER_TABLE_RX_DEF];
+ table->id = EFX_FARCH_FILTER_TABLE_RX_DEF;
+ table->size = EFX_FARCH_FILTER_SIZE_RX_DEF;
- table = &state->table[EFX_FILTER_TABLE_TX_MAC];
- table->id = EFX_FILTER_TABLE_TX_MAC;
+ table = &state->table[EFX_FARCH_FILTER_TABLE_TX_MAC];
+ table->id = EFX_FARCH_FILTER_TABLE_TX_MAC;
table->offset = FR_CZ_TX_MAC_FILTER_TBL0;
table->size = FR_CZ_TX_MAC_FILTER_TBL0_ROWS;
table->step = FR_CZ_TX_MAC_FILTER_TBL0_STEP;
}
- for (table_id = 0; table_id < EFX_FILTER_TABLE_COUNT; table_id++) {
+ for (table_id = 0; table_id < EFX_FARCH_FILTER_TABLE_COUNT; table_id++) {
table = &state->table[table_id];
if (table->size == 0)
continue;
@@ -1078,14 +1111,14 @@ int efx_probe_filters(struct efx_nic *efx)
goto fail;
}
- if (state->table[EFX_FILTER_TABLE_RX_DEF].size) {
+ if (state->table[EFX_FARCH_FILTER_TABLE_RX_DEF].size) {
/* RX default filters must always exist */
unsigned i;
- for (i = 0; i < EFX_FILTER_SIZE_RX_DEF; i++)
- efx_filter_reset_rx_def(efx, i);
+ for (i = 0; i < EFX_FARCH_FILTER_SIZE_RX_DEF; i++)
+ efx_farch_filter_reset_rx_def(efx, i);
}
- efx_filter_push_rx_config(efx);
+ efx_farch_filter_push_rx_config(efx);
return 0;
@@ -1097,9 +1130,9 @@ fail:
void efx_remove_filters(struct efx_nic *efx)
{
struct efx_filter_state *state = efx->filter_state;
- enum efx_filter_table_id table_id;
+ enum efx_farch_filter_table_id table_id;
- for (table_id = 0; table_id < EFX_FILTER_TABLE_COUNT; table_id++) {
+ for (table_id = 0; table_id < EFX_FARCH_FILTER_TABLE_COUNT; table_id++) {
kfree(state->table[table_id].used_bitmap);
vfree(state->table[table_id].spec);
}
@@ -1113,15 +1146,15 @@ void efx_remove_filters(struct efx_nic *efx)
void efx_filter_update_rx_scatter(struct efx_nic *efx)
{
struct efx_filter_state *state = efx->filter_state;
- enum efx_filter_table_id table_id;
- struct efx_filter_table *table;
+ enum efx_farch_filter_table_id table_id;
+ struct efx_farch_filter_table *table;
efx_oword_t filter;
unsigned int filter_idx;
spin_lock_bh(&state->lock);
- for (table_id = EFX_FILTER_TABLE_RX_IP;
- table_id <= EFX_FILTER_TABLE_RX_DEF;
+ for (table_id = EFX_FARCH_FILTER_TABLE_RX_IP;
+ table_id <= EFX_FARCH_FILTER_TABLE_RX_DEF;
table_id++) {
table = &state->table[table_id];
@@ -1138,17 +1171,17 @@ void efx_filter_update_rx_scatter(struct efx_nic *efx)
table->spec[filter_idx].flags &=
~EFX_FILTER_FLAG_RX_SCATTER;
- if (table_id == EFX_FILTER_TABLE_RX_DEF)
- /* Pushed by efx_filter_push_rx_config() */
+ if (table_id == EFX_FARCH_FILTER_TABLE_RX_DEF)
+ /* Pushed by efx_farch_filter_push_rx_config() */
continue;
- efx_filter_build(&filter, &table->spec[filter_idx]);
+ efx_farch_filter_build(&filter, &table->spec[filter_idx]);
efx_writeo(efx, &filter,
table->offset + table->step * filter_idx);
}
}
- efx_filter_push_rx_config(efx);
+ efx_farch_filter_push_rx_config(efx);
spin_unlock_bh(&state->lock);
}
@@ -1222,7 +1255,8 @@ int efx_filter_rfs(struct net_device *net_dev, const struct sk_buff *skb,
bool __efx_filter_rfs_expire(struct efx_nic *efx, unsigned quota)
{
struct efx_filter_state *state = efx->filter_state;
- struct efx_filter_table *table = &state->table[EFX_FILTER_TABLE_RX_IP];
+ struct efx_farch_filter_table *table =
+ &state->table[EFX_FARCH_FILTER_TABLE_RX_IP];
unsigned mask = table->size - 1;
unsigned index;
unsigned stop;
@@ -1242,14 +1276,14 @@ bool __efx_filter_rfs_expire(struct efx_nic *efx, unsigned quota)
netif_info(efx, rx_status, efx->net_dev,
"expiring filter %d [flow %u]\n",
index, state->rps_flow_id[index]);
- efx_filter_table_clear_entry(efx, table, index);
+ efx_farch_filter_table_clear_entry(efx, table, index);
}
index = (index + 1) & mask;
}
state->rps_expire_index = stop;
if (table->used == 0)
- efx_filter_table_reset_search_depth(table);
+ efx_farch_filter_table_reset_search_depth(table);
spin_unlock_bh(&state->lock);
return true;
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* Re: Netfilter owner match breakage
From: Eric Dumazet @ 2013-08-27 20:43 UTC (permalink / raw)
To: Phil Oester; +Cc: netdev, netfilter-devel, edumazet
In-Reply-To: <20130827202445.GA3114@linuxace.com>
On Tue, 2013-08-27 at 13:24 -0700, Phil Oester wrote:
> In commit 90ba9b19 (tcp: tcp_make_synack() can use alloc_skb()), Eric changed
> the call to sock_wmalloc in tcp_make_synack to alloc_skb. In doing so,
> the netfilter owner match lost its ability to block the SYNACK packet on
> outbound listening sockets. For example:
>
> [phil@f19_main ~]$ id
> uid=1000(phil) gid=1000(phil) groups=1000(phil)
>
> [phil@f19_main ~]$ sudo iptables -A OUTPUT -p tcp -m owner --uid-owner 1000 -j REJECT
>
Oh well.
> [phil@f19_main ~]$ echo hi | nc -l -p 8888
>
> Before this commit, attempting to connect to the port 8888 listener generated
> this:
>
> 10.10.10.1.47457 > 10.10.10.110.8888: Flags [S], seq 1855270582 [...]
> 10.10.10.110 > 10.10.10.110: ICMP 10.10.10.1 tcp port 47457 unreachable, length 68
>
Hmm... I think TCP stack should send more SYNACK, no ?
It sounds more logical to block the incoming SYN, but whatever.
> After this commit, the session is established but the first packet of the
> session gets rejected:
>
> 10.10.10.1.47453 > 10.10.10.110.8888: Flags [S], seq 2089355862 [...]
> 10.10.10.110.8888 > 10.10.10.1.47453: Flags [S.], seq 2554257713, ack 2089355863 [...]
> 10.10.10.1.47453 > 10.10.10.110.8888: Flags [.], ack 1, win 115 [...]
> 10.10.10.110 > 10.10.10.110: ICMP 10.10.10.1 tcp port 47453 unreachable, length 63
>
> Reverting 90ba9b19 (and adjusting for the removal of s_data_desired) fixes the
> problem. Is there a better way to do this?
Well, a revert seems OK to me, it was not a critical change.
^ permalink raw reply
* [PATCH net-next 04/16] sfc: Remove unused filter_flags variables and efx_farch_filter_id_flags()
From: Ben Hutchings @ 2013-08-27 20:39 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377635844.13272.69.camel@bwh-desktop.uk.level5networks.com>
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/filter.c | 16 ----------------
1 file changed, 16 deletions(-)
diff --git a/drivers/net/ethernet/sfc/filter.c b/drivers/net/ethernet/sfc/filter.c
index fb01fdb..a588c95 100644
--- a/drivers/net/ethernet/sfc/filter.c
+++ b/drivers/net/ethernet/sfc/filter.c
@@ -609,16 +609,6 @@ static inline unsigned int efx_filter_id_index(u32 id)
return id & EFX_FILTER_INDEX_MASK;
}
-static inline u8 efx_filter_id_flags(u32 id)
-{
- unsigned int range = id >> EFX_FILTER_INDEX_WIDTH;
-
- if (range < EFX_FILTER_MATCH_PRI_COUNT)
- return EFX_FILTER_FLAG_RX;
- else
- return EFX_FILTER_FLAG_TX;
-}
-
u32 efx_filter_get_rx_id_limit(struct efx_nic *efx)
{
struct efx_filter_state *state = efx->filter_state;
@@ -829,7 +819,6 @@ int efx_filter_remove_id_safe(struct efx_nic *efx,
struct efx_filter_table *table;
unsigned int filter_idx;
struct efx_filter_spec *spec;
- u8 filter_flags;
int rc;
table_id = efx_filter_id_table_id(filter_id);
@@ -842,8 +831,6 @@ int efx_filter_remove_id_safe(struct efx_nic *efx,
return -ENOENT;
spec = &table->spec[filter_idx];
- filter_flags = efx_filter_id_flags(filter_id);
-
spin_lock_bh(&state->lock);
if (test_bit(filter_idx, table->used_bitmap) &&
@@ -880,7 +867,6 @@ int efx_filter_get_filter_safe(struct efx_nic *efx,
struct efx_filter_table *table;
struct efx_filter_spec *spec;
unsigned int filter_idx;
- u8 filter_flags;
int rc;
table_id = efx_filter_id_table_id(filter_id);
@@ -893,8 +879,6 @@ int efx_filter_get_filter_safe(struct efx_nic *efx,
return -ENOENT;
spec = &table->spec[filter_idx];
- filter_flags = efx_filter_id_flags(filter_id);
-
spin_lock_bh(&state->lock);
if (test_bit(filter_idx, table->used_bitmap) &&
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [PATCH net-next 03/16] sfc: Do not assume efx_nic_type::ev_fini is idempotent
From: Ben Hutchings @ 2013-08-27 20:39 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377635844.13272.69.camel@bwh-desktop.uk.level5networks.com>
efx_fini_eventq() needs to be idempotent but EF10 firmware is
picky about queue states.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/efx.c | 5 +++++
drivers/net/ethernet/sfc/net_driver.h | 2 ++
2 files changed, 7 insertions(+)
diff --git a/drivers/net/ethernet/sfc/efx.c b/drivers/net/ethernet/sfc/efx.c
index 3977926..49d06ca 100644
--- a/drivers/net/ethernet/sfc/efx.c
+++ b/drivers/net/ethernet/sfc/efx.c
@@ -339,6 +339,7 @@ static void efx_init_eventq(struct efx_channel *channel)
channel->eventq_read_ptr = 0;
efx_nic_init_eventq(channel);
+ channel->eventq_init = true;
}
/* Enable event queue processing and NAPI */
@@ -367,10 +368,14 @@ static void efx_stop_eventq(struct efx_channel *channel)
static void efx_fini_eventq(struct efx_channel *channel)
{
+ if (!channel->eventq_init)
+ return;
+
netif_dbg(channel->efx, drv, channel->efx->net_dev,
"chan %d fini event queue\n", channel->channel);
efx_nic_fini_eventq(channel);
+ channel->eventq_init = false;
}
static void efx_remove_eventq(struct efx_channel *channel)
diff --git a/drivers/net/ethernet/sfc/net_driver.h b/drivers/net/ethernet/sfc/net_driver.h
index 7aa0e0f..694c572 100644
--- a/drivers/net/ethernet/sfc/net_driver.h
+++ b/drivers/net/ethernet/sfc/net_driver.h
@@ -356,6 +356,7 @@ enum efx_rx_alloc_method {
* @efx: Associated Efx NIC
* @channel: Channel instance number
* @type: Channel type definition
+ * @eventq_init: Event queue initialised flag
* @enabled: Channel enabled indicator
* @irq: IRQ number (MSI and MSI-X only)
* @irq_moderation: IRQ moderation value (in hardware ticks)
@@ -387,6 +388,7 @@ struct efx_channel {
struct efx_nic *efx;
int channel;
const struct efx_channel_type *type;
+ bool eventq_init;
bool enabled;
int irq;
unsigned int irq_moderation;
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [PATCH net-next 02/16] sfc: EFX_WORKAROUND_ALWAYS is really specific to Falcon-architecture
From: Ben Hutchings @ 2013-08-27 20:38 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377635844.13272.69.camel@bwh-desktop.uk.level5networks.com>
The workarounds that currently use EFX_WORKAROUND_ALWAYS are in
Falcon-specific or Falcon-arch-specific code, so get rid of the
conditions altogether. Add/move comments as appropriate.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/falcon.c | 35 ++++++++++++++--------------------
drivers/net/ethernet/sfc/farch.c | 10 +++++-----
drivers/net/ethernet/sfc/workarounds.h | 10 ----------
3 files changed, 19 insertions(+), 36 deletions(-)
diff --git a/drivers/net/ethernet/sfc/falcon.c b/drivers/net/ethernet/sfc/falcon.c
index 17bd7dc..0fd8a88 100644
--- a/drivers/net/ethernet/sfc/falcon.c
+++ b/drivers/net/ethernet/sfc/falcon.c
@@ -686,7 +686,7 @@ static void falcon_ack_status_intr(struct efx_nic *efx)
return;
/* We expect xgmii faults if the wireside link is down */
- if (!EFX_WORKAROUND_5147(efx) || !efx->link_state.up)
+ if (!efx->link_state.up)
return;
/* We can only use this interrupt to signal the negative edge of
@@ -795,29 +795,22 @@ static void falcon_reconfigure_xgxs_core(struct efx_nic *efx)
bool xgxs_loopback = (efx->loopback_mode == LOOPBACK_XGXS);
bool xaui_loopback = (efx->loopback_mode == LOOPBACK_XAUI);
bool xgmii_loopback = (efx->loopback_mode == LOOPBACK_XGMII);
+ bool old_xgmii_loopback, old_xgxs_loopback, old_xaui_loopback;
/* XGXS block is flaky and will need to be reset if moving
* into our out of XGMII, XGXS or XAUI loopbacks. */
- if (EFX_WORKAROUND_5147(efx)) {
- bool old_xgmii_loopback, old_xgxs_loopback, old_xaui_loopback;
- bool reset_xgxs;
-
- efx_reado(efx, ®, FR_AB_XX_CORE_STAT);
- old_xgxs_loopback = EFX_OWORD_FIELD(reg, FRF_AB_XX_XGXS_LB_EN);
- old_xgmii_loopback =
- EFX_OWORD_FIELD(reg, FRF_AB_XX_XGMII_LB_EN);
-
- efx_reado(efx, ®, FR_AB_XX_SD_CTL);
- old_xaui_loopback = EFX_OWORD_FIELD(reg, FRF_AB_XX_LPBKA);
+ efx_reado(efx, ®, FR_AB_XX_CORE_STAT);
+ old_xgxs_loopback = EFX_OWORD_FIELD(reg, FRF_AB_XX_XGXS_LB_EN);
+ old_xgmii_loopback = EFX_OWORD_FIELD(reg, FRF_AB_XX_XGMII_LB_EN);
- /* The PHY driver may have turned XAUI off */
- reset_xgxs = ((xgxs_loopback != old_xgxs_loopback) ||
- (xaui_loopback != old_xaui_loopback) ||
- (xgmii_loopback != old_xgmii_loopback));
+ efx_reado(efx, ®, FR_AB_XX_SD_CTL);
+ old_xaui_loopback = EFX_OWORD_FIELD(reg, FRF_AB_XX_LPBKA);
- if (reset_xgxs)
- falcon_reset_xaui(efx);
- }
+ /* The PHY driver may have turned XAUI off */
+ if ((xgxs_loopback != old_xgxs_loopback) ||
+ (xaui_loopback != old_xaui_loopback) ||
+ (xgmii_loopback != old_xgmii_loopback))
+ falcon_reset_xaui(efx);
efx_reado(efx, ®, FR_AB_XX_CORE_STAT);
EFX_SET_OWORD_FIELD(reg, FRF_AB_XX_FORCE_SIG,
@@ -946,8 +939,8 @@ static void falcon_poll_xmac(struct efx_nic *efx)
{
struct falcon_nic_data *nic_data = efx->nic_data;
- if (!EFX_WORKAROUND_5147(efx) || !efx->link_state.up ||
- !nic_data->xmac_poll_required)
+ /* We expect xgmii faults if the wireside link is down */
+ if (!efx->link_state.up || !nic_data->xmac_poll_required)
return;
nic_data->xmac_poll_required = !falcon_xmac_link_ok_retry(efx, 1);
diff --git a/drivers/net/ethernet/sfc/farch.c b/drivers/net/ethernet/sfc/farch.c
index c3d07c5..7f50882 100644
--- a/drivers/net/ethernet/sfc/farch.c
+++ b/drivers/net/ethernet/sfc/farch.c
@@ -830,8 +830,7 @@ efx_farch_handle_tx_event(struct efx_channel *channel, efx_qword_t *event)
netif_tx_lock(efx->net_dev);
efx_farch_notify_tx_desc(tx_queue);
netif_tx_unlock(efx->net_dev);
- } else if (EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_PKT_ERR) &&
- EFX_WORKAROUND_10727(efx)) {
+ } else if (EFX_QWORD_FIELD(*event, FSF_AZ_TX_EV_PKT_ERR)) {
efx_schedule_reset(efx, RESET_TYPE_TX_DESC_FETCH);
} else {
netif_err(efx, tx_err, efx->net_dev,
@@ -1531,8 +1530,7 @@ irqreturn_t efx_farch_legacy_interrupt(int irq, void *dev_id)
}
if (queues != 0) {
- if (EFX_WORKAROUND_15783(efx))
- efx->irq_zero_count = 0;
+ efx->irq_zero_count = 0;
/* Schedule processing of any interrupting queues */
if (likely(soft_enabled)) {
@@ -1544,9 +1542,11 @@ irqreturn_t efx_farch_legacy_interrupt(int irq, void *dev_id)
}
result = IRQ_HANDLED;
- } else if (EFX_WORKAROUND_15783(efx)) {
+ } else {
efx_qword_t *event;
+ /* Legacy ISR read can return zero once (SF bug 15783) */
+
/* We can't return IRQ_HANDLED more than once on seeing ISR=0
* because this might be a shared interrupt. */
if (efx->irq_zero_count++ == 0)
diff --git a/drivers/net/ethernet/sfc/workarounds.h b/drivers/net/ethernet/sfc/workarounds.h
index dff565a..7e5be1d 100644
--- a/drivers/net/ethernet/sfc/workarounds.h
+++ b/drivers/net/ethernet/sfc/workarounds.h
@@ -15,25 +15,15 @@
* Bug numbers are from Solarflare's Bugzilla.
*/
-#define EFX_WORKAROUND_ALWAYS(efx) 1
#define EFX_WORKAROUND_FALCON_A(efx) (efx_nic_rev(efx) <= EFX_REV_FALCON_A1)
#define EFX_WORKAROUND_FALCON_AB(efx) (efx_nic_rev(efx) <= EFX_REV_FALCON_B0)
#define EFX_WORKAROUND_SIENA(efx) (efx_nic_rev(efx) == EFX_REV_SIENA_A0)
#define EFX_WORKAROUND_10G(efx) 1
-/* XAUI resets if link not detected */
-#define EFX_WORKAROUND_5147 EFX_WORKAROUND_ALWAYS
-/* RX PCIe double split performance issue */
-#define EFX_WORKAROUND_7575 EFX_WORKAROUND_ALWAYS
/* Bit-bashed I2C reads cause performance drop */
#define EFX_WORKAROUND_7884 EFX_WORKAROUND_10G
-/* TX_EV_PKT_ERR can be caused by a dangling TX descriptor
- * or a PCIe error (bug 11028) */
-#define EFX_WORKAROUND_10727 EFX_WORKAROUND_ALWAYS
/* Truncated IPv4 packets can confuse the TX packet parser */
#define EFX_WORKAROUND_15592 EFX_WORKAROUND_FALCON_AB
-/* Legacy ISR read can return zero once */
-#define EFX_WORKAROUND_15783 EFX_WORKAROUND_ALWAYS
/* Legacy interrupt storm when interrupt fifo fills */
#define EFX_WORKAROUND_17213 EFX_WORKAROUND_SIENA
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* [PATCH net-next 01/16] sfc: Get rid of per-NIC-type phys_addr_channels and mem_map_size
From: Ben Hutchings @ 2013-08-27 20:38 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
In-Reply-To: <1377635844.13272.69.camel@bwh-desktop.uk.level5networks.com>
EF10 functions don't have a fixed BAR size, and the minimum is not
large enough for all the queues we might want to allocate. We have to
find out the BAR size at run-time, and therefore phys_addr_channels
and mem_map_size cannot be defined per-NIC-type.
Change efx_nic_type::mem_map_size to a function pointer which is
called to find the wanted memory map size (before probe).
Replace efx_nic_type::phys_addr_channels with efx_nic::max_channels,
to be initialised by the probe function.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
---
drivers/net/ethernet/sfc/efx.c | 17 ++++++-----------
drivers/net/ethernet/sfc/falcon.c | 29 ++++++++++++++++++-----------
drivers/net/ethernet/sfc/net_driver.h | 9 ++++-----
drivers/net/ethernet/sfc/siena.c | 14 +++++++++-----
4 files changed, 37 insertions(+), 32 deletions(-)
diff --git a/drivers/net/ethernet/sfc/efx.c b/drivers/net/ethernet/sfc/efx.c
index 872b9f5..3977926 100644
--- a/drivers/net/ethernet/sfc/efx.c
+++ b/drivers/net/ethernet/sfc/efx.c
@@ -1084,6 +1084,7 @@ static int efx_init_io(struct efx_nic *efx)
{
struct pci_dev *pci_dev = efx->pci_dev;
dma_addr_t dma_mask = efx->type->max_dma_mask;
+ unsigned int mem_map_size = efx->type->mem_map_size(efx);
int rc;
netif_dbg(efx, probe, efx->net_dev, "initialising I/O\n");
@@ -1136,20 +1137,18 @@ static int efx_init_io(struct efx_nic *efx)
rc = -EIO;
goto fail3;
}
- efx->membase = ioremap_nocache(efx->membase_phys,
- efx->type->mem_map_size);
+ efx->membase = ioremap_nocache(efx->membase_phys, mem_map_size);
if (!efx->membase) {
netif_err(efx, probe, efx->net_dev,
"could not map memory BAR at %llx+%x\n",
- (unsigned long long)efx->membase_phys,
- efx->type->mem_map_size);
+ (unsigned long long)efx->membase_phys, mem_map_size);
rc = -ENOMEM;
goto fail4;
}
netif_dbg(efx, probe, efx->net_dev,
"memory BAR at %llx+%x (virtual %p)\n",
- (unsigned long long)efx->membase_phys,
- efx->type->mem_map_size, efx->membase);
+ (unsigned long long)efx->membase_phys, mem_map_size,
+ efx->membase);
return 0;
@@ -1228,8 +1227,6 @@ static unsigned int efx_wanted_parallelism(struct efx_nic *efx)
*/
static int efx_probe_interrupts(struct efx_nic *efx)
{
- unsigned int max_channels =
- min(efx->type->phys_addr_channels, EFX_MAX_CHANNELS);
unsigned int extra_channels = 0;
unsigned int i, j;
int rc;
@@ -1246,7 +1243,7 @@ static int efx_probe_interrupts(struct efx_nic *efx)
if (separate_tx_channels)
n_channels *= 2;
n_channels += extra_channels;
- n_channels = min(n_channels, max_channels);
+ n_channels = min(n_channels, efx->max_channels);
for (i = 0; i < n_channels; i++)
xentries[i].entry = i;
@@ -2489,8 +2486,6 @@ static int efx_init_struct(struct efx_nic *efx,
efx->msi_context[i].index = i;
}
- EFX_BUG_ON_PARANOID(efx->type->phys_addr_channels > EFX_MAX_CHANNELS);
-
/* Higher numbered interrupt modes are less capable! */
efx->interrupt_mode = max(efx->type->max_interrupt_mode,
interrupt_mode);
diff --git a/drivers/net/ethernet/sfc/falcon.c b/drivers/net/ethernet/sfc/falcon.c
index e556a5d..17bd7dc 100644
--- a/drivers/net/ethernet/sfc/falcon.c
+++ b/drivers/net/ethernet/sfc/falcon.c
@@ -1970,6 +1970,20 @@ static void falcon_probe_spi_devices(struct efx_nic *efx)
large_eeprom_type);
}
+static unsigned int falcon_a1_mem_map_size(struct efx_nic *efx)
+{
+ return 0x20000;
+}
+
+static unsigned int falcon_b0_mem_map_size(struct efx_nic *efx)
+{
+ /* Map everything up to and including the RSS indirection table.
+ * The PCI core takes care of mapping the MSI-X tables.
+ */
+ return FR_BZ_RX_INDIRECTION_TBL +
+ FR_BZ_RX_INDIRECTION_TBL_STEP * FR_BZ_RX_INDIRECTION_TBL_ROWS;
+}
+
static int falcon_probe_nic(struct efx_nic *efx)
{
struct falcon_nic_data *nic_data;
@@ -2060,6 +2074,8 @@ static int falcon_probe_nic(struct efx_nic *efx)
goto fail5;
}
+ efx->max_channels = (efx_nic_rev(efx) <= EFX_REV_FALCON_A1 ? 4 :
+ EFX_MAX_CHANNELS);
efx->timer_quantum_ns = 4968; /* 621 cycles */
/* Initialise I2C adapter */
@@ -2339,6 +2355,7 @@ static int falcon_set_wol(struct efx_nic *efx, u32 type)
*/
const struct efx_nic_type falcon_a1_nic_type = {
+ .mem_map_size = falcon_a1_mem_map_size,
.probe = falcon_probe_nic,
.remove = falcon_remove_nic,
.init = falcon_init_nic,
@@ -2391,7 +2408,6 @@ const struct efx_nic_type falcon_a1_nic_type = {
.ev_test_generate = efx_farch_ev_test_generate,
.revision = EFX_REV_FALCON_A1,
- .mem_map_size = 0x20000,
.txd_ptr_tbl_base = FR_AA_TX_DESC_PTR_TBL_KER,
.rxd_ptr_tbl_base = FR_AA_RX_DESC_PTR_TBL_KER,
.buf_tbl_base = FR_AA_BUF_FULL_TBL_KER,
@@ -2401,13 +2417,13 @@ const struct efx_nic_type falcon_a1_nic_type = {
.rx_buffer_padding = 0x24,
.can_rx_scatter = false,
.max_interrupt_mode = EFX_INT_MODE_MSI,
- .phys_addr_channels = 4,
.timer_period_max = 1 << FRF_AB_TC_TIMER_VAL_WIDTH,
.offload_features = NETIF_F_IP_CSUM,
.mcdi_max_ver = -1,
};
const struct efx_nic_type falcon_b0_nic_type = {
+ .mem_map_size = falcon_b0_mem_map_size,
.probe = falcon_probe_nic,
.remove = falcon_remove_nic,
.init = falcon_init_nic,
@@ -2461,12 +2477,6 @@ const struct efx_nic_type falcon_b0_nic_type = {
.ev_test_generate = efx_farch_ev_test_generate,
.revision = EFX_REV_FALCON_B0,
- /* Map everything up to and including the RSS indirection
- * table. Don't map MSI-X table, MSI-X PBA since Linux
- * requires that they not be mapped. */
- .mem_map_size = (FR_BZ_RX_INDIRECTION_TBL +
- FR_BZ_RX_INDIRECTION_TBL_STEP *
- FR_BZ_RX_INDIRECTION_TBL_ROWS),
.txd_ptr_tbl_base = FR_BZ_TX_DESC_PTR_TBL,
.rxd_ptr_tbl_base = FR_BZ_RX_DESC_PTR_TBL,
.buf_tbl_base = FR_BZ_BUF_FULL_TBL,
@@ -2477,9 +2487,6 @@ const struct efx_nic_type falcon_b0_nic_type = {
.rx_buffer_padding = 0,
.can_rx_scatter = true,
.max_interrupt_mode = EFX_INT_MODE_MSIX,
- .phys_addr_channels = 32, /* Hardware limit is 64, but the legacy
- * interrupt handler only supports 32
- * channels */
.timer_period_max = 1 << FRF_AB_TC_TIMER_VAL_WIDTH,
.offload_features = NETIF_F_IP_CSUM | NETIF_F_RXHASH | NETIF_F_NTUPLE,
.mcdi_max_ver = -1,
diff --git a/drivers/net/ethernet/sfc/net_driver.h b/drivers/net/ethernet/sfc/net_driver.h
index e1deec4..7aa0e0f 100644
--- a/drivers/net/ethernet/sfc/net_driver.h
+++ b/drivers/net/ethernet/sfc/net_driver.h
@@ -832,6 +832,8 @@ struct efx_nic {
unsigned rx_dc_base;
unsigned sram_lim_qw;
unsigned next_buffer_table;
+
+ unsigned int max_channels;
unsigned n_channels;
unsigned n_rx_channels;
unsigned rss_spread;
@@ -939,6 +941,7 @@ static inline unsigned int efx_port_num(struct efx_nic *efx)
/**
* struct efx_nic_type - Efx device type definition
+ * @mem_map_size: Get memory BAR mapped size
* @probe: Probe the controller
* @remove: Free resources allocated by probe()
* @init: Initialise the controller
@@ -1012,7 +1015,6 @@ static inline unsigned int efx_port_num(struct efx_nic *efx)
* @ev_read_ack: Acknowledge read events on a queue, rearming its IRQ
* @ev_test_generate: Generate a test event
* @revision: Hardware architecture revision
- * @mem_map_size: Memory BAR mapped size
* @txd_ptr_tbl_base: TX descriptor ring base address
* @rxd_ptr_tbl_base: RX descriptor ring base address
* @buf_tbl_base: Buffer table base address
@@ -1024,14 +1026,13 @@ static inline unsigned int efx_port_num(struct efx_nic *efx)
* @can_rx_scatter: NIC is able to scatter packet to multiple buffers
* @max_interrupt_mode: Highest capability interrupt mode supported
* from &enum efx_init_mode.
- * @phys_addr_channels: Number of channels with physically addressed
- * descriptors
* @timer_period_max: Maximum period of interrupt timer (in ticks)
* @offload_features: net_device feature flags for protocol offload
* features implemented in hardware
* @mcdi_max_ver: Maximum MCDI version supported
*/
struct efx_nic_type {
+ unsigned int (*mem_map_size)(struct efx_nic *efx);
int (*probe)(struct efx_nic *efx);
void (*remove)(struct efx_nic *efx);
int (*init)(struct efx_nic *efx);
@@ -1092,7 +1093,6 @@ struct efx_nic_type {
void (*ev_test_generate)(struct efx_channel *channel);
int revision;
- unsigned int mem_map_size;
unsigned int txd_ptr_tbl_base;
unsigned int rxd_ptr_tbl_base;
unsigned int buf_tbl_base;
@@ -1103,7 +1103,6 @@ struct efx_nic_type {
unsigned int rx_buffer_padding;
bool can_rx_scatter;
unsigned int max_interrupt_mode;
- unsigned int phys_addr_channels;
unsigned int timer_period_max;
netdev_features_t offload_features;
int mcdi_max_ver;
diff --git a/drivers/net/ethernet/sfc/siena.c b/drivers/net/ethernet/sfc/siena.c
index fee0d2d..23e5731 100644
--- a/drivers/net/ethernet/sfc/siena.c
+++ b/drivers/net/ethernet/sfc/siena.c
@@ -187,6 +187,12 @@ static void siena_dimension_resources(struct efx_nic *efx)
efx_farch_dimension_resources(efx, FR_CZ_BUF_FULL_TBL_ROWS / 2);
}
+static unsigned int siena_mem_map_size(struct efx_nic *efx)
+{
+ return FR_CZ_MC_TREG_SMEM +
+ FR_CZ_MC_TREG_SMEM_STEP * FR_CZ_MC_TREG_SMEM_ROWS;
+}
+
static int siena_probe_nic(struct efx_nic *efx)
{
struct siena_nic_data *nic_data;
@@ -207,6 +213,8 @@ static int siena_probe_nic(struct efx_nic *efx)
goto fail1;
}
+ efx->max_channels = EFX_MAX_CHANNELS;
+
efx_reado(efx, ®, FR_AZ_CS_DEBUG);
efx->port_num = EFX_OWORD_FIELD(reg, FRF_CZ_CS_PORT_NUM) - 1;
@@ -670,6 +678,7 @@ static int siena_mcdi_poll_reboot(struct efx_nic *efx)
*/
const struct efx_nic_type siena_a0_nic_type = {
+ .mem_map_size = siena_mem_map_size,
.probe = siena_probe_nic,
.remove = siena_remove_nic,
.init = siena_init_nic,
@@ -729,8 +738,6 @@ const struct efx_nic_type siena_a0_nic_type = {
.ev_test_generate = efx_farch_ev_test_generate,
.revision = EFX_REV_SIENA_A0,
- .mem_map_size = (FR_CZ_MC_TREG_SMEM +
- FR_CZ_MC_TREG_SMEM_STEP * FR_CZ_MC_TREG_SMEM_ROWS),
.txd_ptr_tbl_base = FR_BZ_TX_DESC_PTR_TBL,
.rxd_ptr_tbl_base = FR_BZ_RX_DESC_PTR_TBL,
.buf_tbl_base = FR_BZ_BUF_FULL_TBL,
@@ -741,9 +748,6 @@ const struct efx_nic_type siena_a0_nic_type = {
.rx_buffer_padding = 0,
.can_rx_scatter = true,
.max_interrupt_mode = EFX_INT_MODE_MSIX,
- .phys_addr_channels = 32, /* Hardware limit is 64, but the legacy
- * interrupt handler only supports 32
- * channels */
.timer_period_max = 1 << FRF_CZ_TC_TIMER_VAL_WIDTH,
.offload_features = (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
NETIF_F_RXHASH | NETIF_F_NTUPLE),
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply related
* Pull request: sfc-next 2013-08-27
From: Ben Hutchings @ 2013-08-27 20:37 UTC (permalink / raw)
To: David Miller; +Cc: netdev, linux-net-drivers
The following changes since commit f76fe120d81c96fa2a17ae41f0647c963dbb43cd:
sfc: Update and improve kernel-doc for efx_mcdi_state & efx_mcdi_iface (2013-08-21 20:20:41 +0100)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/bwh/sfc-next.git for-davem
for you to fetch changes up to b766630b351c68c0383831dba9b81a905e5e84c6:
sfc: Eliminate struct efx_mtd (2013-08-22 19:26:04 +0100)
More refactoring and cleanup, particularly around filter management.
Ben.
----------------------------------------------------------------
Ben Hutchings (16):
sfc: Get rid of per-NIC-type phys_addr_channels and mem_map_size
sfc: EFX_WORKAROUND_ALWAYS is really specific to Falcon-architecture
sfc: Do not assume efx_nic_type::ev_fini is idempotent
sfc: Remove unused filter_flags variables and efx_farch_filter_id_flags()
sfc: Rename Falcon-arch filter implementation types and functions
sfc: Name the RX drop queue ID
sfc: Extend and abstract efx_filter_spec to cover Huntington/EF10
sfc: Split Falcon-arch-specific and common filter state
sfc: Refactor Falcon-arch search limit reset
sfc: Make most filter operations NIC-type-specific
sfc: Refactor Falcon-arch filter removal
sfc: Add flag for stack-owned RX MAC filters
sfc: Define and use MCDI_POPULATE_DWORD_{1,2,3,4,5,6,7}
sfc: Cleanup Falcon-arch simple MAC filter state
sfc: Rename SPI stuff to show that it is Falcon-specific
sfc: Eliminate struct efx_mtd
drivers/net/ethernet/sfc/Makefile | 1 -
drivers/net/ethernet/sfc/efx.c | 96 +--
drivers/net/ethernet/sfc/efx.h | 107 ++-
drivers/net/ethernet/sfc/ethtool.c | 215 +++---
drivers/net/ethernet/sfc/falcon.c | 131 ++--
drivers/net/ethernet/sfc/farch.c | 1171 ++++++++++++++++++++++++++++-
drivers/net/ethernet/sfc/filter.c | 1274 --------------------------------
drivers/net/ethernet/sfc/filter.h | 234 ++++--
drivers/net/ethernet/sfc/mcdi.h | 54 ++
drivers/net/ethernet/sfc/mcdi_port.c | 29 +-
drivers/net/ethernet/sfc/mtd.c | 307 ++++----
drivers/net/ethernet/sfc/net_driver.h | 79 +-
drivers/net/ethernet/sfc/nic.h | 33 +-
drivers/net/ethernet/sfc/rx.c | 94 +++
drivers/net/ethernet/sfc/siena.c | 32 +-
drivers/net/ethernet/sfc/spi.h | 18 +-
drivers/net/ethernet/sfc/workarounds.h | 10 -
17 files changed, 2118 insertions(+), 1767 deletions(-)
delete mode 100644 drivers/net/ethernet/sfc/filter.c
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply
* [PATCH v2] fec: Use NAPI_POLL_WEIGHT
From: Fabio Estevam @ 2013-08-27 20:35 UTC (permalink / raw)
To: davem; +Cc: netdev, Fabio Estevam
From: Fabio Estevam <fabio.estevam@freescale.com>
Instead of using a custom 'FEC_NAPI_WEIGHT', just use the generic
'NAPI_POLL_WEIGHT' definition instead.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
---
Changes since v1:
- Fix typo in commit log (NAPI_POLL_WEIGHT was written twice in v1)
drivers/net/ethernet/freescale/fec_main.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/freescale/fec_main.c b/drivers/net/ethernet/freescale/fec_main.c
index fdf9307..c88a129 100644
--- a/drivers/net/ethernet/freescale/fec_main.c
+++ b/drivers/net/ethernet/freescale/fec_main.c
@@ -69,7 +69,6 @@ static void set_multicast_list(struct net_device *ndev);
#endif
#define DRIVER_NAME "fec"
-#define FEC_NAPI_WEIGHT 64
/* Pause frame feild and FIFO threshold */
#define FEC_ENET_FCE (1 << 5)
@@ -1981,7 +1980,7 @@ static int fec_enet_init(struct net_device *ndev)
ndev->ethtool_ops = &fec_enet_ethtool_ops;
writel(FEC_RX_DISABLED_IMASK, fep->hwp + FEC_IMASK);
- netif_napi_add(ndev, &fep->napi, fec_enet_rx_napi, FEC_NAPI_WEIGHT);
+ netif_napi_add(ndev, &fep->napi, fec_enet_rx_napi, NAPI_POLL_WEIGHT);
if (id_entry->driver_data & FEC_QUIRK_HAS_VLAN) {
/* enable hw VLAN support */
--
1.8.1.2
^ permalink raw reply related
* Re: [net-next v2 1/8] i40e: main driver core
From: Nelson, Shannon @ 2013-08-27 20:34 UTC (permalink / raw)
To: Nelson, Shannon, David Miller, Kirsher, Jeffrey T
Cc: e1000-devel@lists.sourceforge.net, netdev@vger.kernel.org,
Brandeburg, Jesse, sassmann@redhat.com, gospo@redhat.com
In-Reply-To: <FC41C24E35F18A40888AACA1A36F3E416C615C3B@FMSMSX102.amr.corp.intel.com>
> -----Original Message-----
> From: David Miller [mailto:davem@davemloft.net]
> Sent: Friday, August 23, 2013 12:28 AM
[...]
>
> > +{
> > + int i;
> > + struct i40e_pf *pf = vsi->back;
>
> Please order local variable declarations from longest line to shortest.
I understand the aesthetics as it does make the code look a little cleaner, and we can do this with a lot of our functions. However, there are several instances where one declaration initialization depends on a previous declaration, and trying to organize by line length breaks these relationships. Do you mind if we're not perfect on following this one?
Also, perhaps I haven't watched closely enough, but I don't remember seeing this comment on other code. Is this a new recommendation that will be going into the CodingStyle document?
Thanks,
sln
------------------------------------------------------------------------------
Learn the latest--Visual Studio 2012, SharePoint 2013, SQL 2012, more!
Discover the easy way to master current and previous Microsoft technologies
and advance your career. Get an incredible 1,500+ hours of step-by-step
tutorial videos with LearnDevNow. Subscribe today and save!
http://pubads.g.doubleclick.net/gampad/clk?id=58040911&iu=/4140/ostg.clktrk
_______________________________________________
E1000-devel mailing list
E1000-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel® Ethernet, visit http://communities.intel.com/community/wired
^ permalink raw reply
* Netfilter owner match breakage
From: Phil Oester @ 2013-08-27 20:24 UTC (permalink / raw)
To: netdev, netfilter-devel; +Cc: edumazet
In commit 90ba9b19 (tcp: tcp_make_synack() can use alloc_skb()), Eric changed
the call to sock_wmalloc in tcp_make_synack to alloc_skb. In doing so,
the netfilter owner match lost its ability to block the SYNACK packet on
outbound listening sockets. For example:
[phil@f19_main ~]$ id
uid=1000(phil) gid=1000(phil) groups=1000(phil)
[phil@f19_main ~]$ sudo iptables -A OUTPUT -p tcp -m owner --uid-owner 1000 -j REJECT
[phil@f19_main ~]$ echo hi | nc -l -p 8888
Before this commit, attempting to connect to the port 8888 listener generated
this:
10.10.10.1.47457 > 10.10.10.110.8888: Flags [S], seq 1855270582 [...]
10.10.10.110 > 10.10.10.110: ICMP 10.10.10.1 tcp port 47457 unreachable, length 68
After this commit, the session is established but the first packet of the
session gets rejected:
10.10.10.1.47453 > 10.10.10.110.8888: Flags [S], seq 2089355862 [...]
10.10.10.110.8888 > 10.10.10.1.47453: Flags [S.], seq 2554257713, ack 2089355863 [...]
10.10.10.1.47453 > 10.10.10.110.8888: Flags [.], ack 1, win 115 [...]
10.10.10.110 > 10.10.10.110: ICMP 10.10.10.1 tcp port 47453 unreachable, length 63
Reverting 90ba9b19 (and adjusting for the removal of s_data_desired) fixes the
problem. Is there a better way to do this?
Phil
^ permalink raw reply
* [PATCH net-next 11/11] openvswitch: optimize flow compare and mask functions
From: Jesse Gross @ 2013-08-27 20:20 UTC (permalink / raw)
To: David Miller; +Cc: netdev, dev, Andy Zhou, Jesse Gross
In-Reply-To: <1377634848-34327-1-git-send-email-jesse@nicira.com>
From: Andy Zhou <azhou@nicira.com>
Make sure the sw_flow_key structure and valid mask boundaries are always
machine word aligned. Optimize the flow compare and mask operations
using machine word size operations. This patch improves throughput on
average by 15% when CPU is the bottleneck of forwarding packets.
This patch is inspired by ideas and code from a patch submitted by Peter
Klausler titled "replace memcmp() with specialized comparator".
However, The original patch only optimizes for architectures
support unaligned machine word access. This patch optimizes for all
architectures.
Signed-off-by: Andy Zhou <azhou@nicira.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>
---
net/openvswitch/flow.c | 64 +++++++++++++++++++++++++++++++++-----------------
net/openvswitch/flow.h | 19 +--------------
2 files changed, 44 insertions(+), 39 deletions(-)
diff --git a/net/openvswitch/flow.c b/net/openvswitch/flow.c
index 80bcb96..ad1aeeb 100644
--- a/net/openvswitch/flow.c
+++ b/net/openvswitch/flow.c
@@ -54,8 +54,8 @@ static void update_range__(struct sw_flow_match *match,
size_t offset, size_t size, bool is_mask)
{
struct sw_flow_key_range *range = NULL;
- size_t start = offset;
- size_t end = offset + size;
+ size_t start = rounddown(offset, sizeof(long));
+ size_t end = roundup(offset + size, sizeof(long));
if (!is_mask)
range = &match->range;
@@ -102,6 +102,11 @@ static void update_range__(struct sw_flow_match *match,
} \
} while (0)
+static u16 range_n_bytes(const struct sw_flow_key_range *range)
+{
+ return range->end - range->start;
+}
+
void ovs_match_init(struct sw_flow_match *match,
struct sw_flow_key *key,
struct sw_flow_mask *mask)
@@ -370,16 +375,17 @@ static bool icmp6hdr_ok(struct sk_buff *skb)
void ovs_flow_key_mask(struct sw_flow_key *dst, const struct sw_flow_key *src,
const struct sw_flow_mask *mask)
{
- u8 *m = (u8 *)&mask->key + mask->range.start;
- u8 *s = (u8 *)src + mask->range.start;
- u8 *d = (u8 *)dst + mask->range.start;
+ const long *m = (long *)((u8 *)&mask->key + mask->range.start);
+ const long *s = (long *)((u8 *)src + mask->range.start);
+ long *d = (long *)((u8 *)dst + mask->range.start);
int i;
- memset(dst, 0, sizeof(*dst));
- for (i = 0; i < ovs_sw_flow_mask_size_roundup(mask); i++) {
- *d = *s & *m;
- d++, s++, m++;
- }
+ /* The memory outside of the 'mask->range' are not set since
+ * further operations on 'dst' only uses contents within
+ * 'mask->range'.
+ */
+ for (i = 0; i < range_n_bytes(&mask->range); i += sizeof(long))
+ *d++ = *s++ & *m++;
}
#define TCP_FLAGS_OFFSET 13
@@ -1000,8 +1006,13 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key)
static u32 ovs_flow_hash(const struct sw_flow_key *key, int key_start,
int key_end)
{
- return jhash2((u32 *)((u8 *)key + key_start),
- DIV_ROUND_UP(key_end - key_start, sizeof(u32)), 0);
+ u32 *hash_key = (u32 *)((u8 *)key + key_start);
+ int hash_u32s = (key_end - key_start) >> 2;
+
+ /* Make sure number of hash bytes are multiple of u32. */
+ BUILD_BUG_ON(sizeof(long) % sizeof(u32));
+
+ return jhash2(hash_key, hash_u32s, 0);
}
static int flow_key_start(const struct sw_flow_key *key)
@@ -1009,17 +1020,25 @@ static int flow_key_start(const struct sw_flow_key *key)
if (key->tun_key.ipv4_dst)
return 0;
else
- return offsetof(struct sw_flow_key, phy);
+ return rounddown(offsetof(struct sw_flow_key, phy),
+ sizeof(long));
}
static bool __cmp_key(const struct sw_flow_key *key1,
const struct sw_flow_key *key2, int key_start, int key_end)
{
- return !memcmp((u8 *)key1 + key_start,
- (u8 *)key2 + key_start, (key_end - key_start));
+ const long *cp1 = (long *)((u8 *)key1 + key_start);
+ const long *cp2 = (long *)((u8 *)key2 + key_start);
+ long diffs = 0;
+ int i;
+
+ for (i = key_start; i < key_end; i += sizeof(long))
+ diffs |= *cp1++ ^ *cp2++;
+
+ return diffs == 0;
}
-static bool __flow_cmp_key(const struct sw_flow *flow,
+static bool __flow_cmp_masked_key(const struct sw_flow *flow,
const struct sw_flow_key *key, int key_start, int key_end)
{
return __cmp_key(&flow->key, key, key_start, key_end);
@@ -1056,7 +1075,7 @@ struct sw_flow *ovs_flow_lookup_unmasked_key(struct flow_table *table,
}
static struct sw_flow *ovs_masked_flow_lookup(struct flow_table *table,
- const struct sw_flow_key *flow_key,
+ const struct sw_flow_key *unmasked,
struct sw_flow_mask *mask)
{
struct sw_flow *flow;
@@ -1066,12 +1085,13 @@ static struct sw_flow *ovs_masked_flow_lookup(struct flow_table *table,
u32 hash;
struct sw_flow_key masked_key;
- ovs_flow_key_mask(&masked_key, flow_key, mask);
+ ovs_flow_key_mask(&masked_key, unmasked, mask);
hash = ovs_flow_hash(&masked_key, key_start, key_end);
head = find_bucket(table, hash);
hlist_for_each_entry_rcu(flow, head, hash_node[table->node_ver]) {
if (flow->mask == mask &&
- __flow_cmp_key(flow, &masked_key, key_start, key_end))
+ __flow_cmp_masked_key(flow, &masked_key,
+ key_start, key_end))
return flow;
}
return NULL;
@@ -1961,6 +1981,8 @@ nla_put_failure:
* Returns zero if successful or a negative error code. */
int ovs_flow_init(void)
{
+ BUILD_BUG_ON(sizeof(struct sw_flow_key) % sizeof(long));
+
flow_cache = kmem_cache_create("sw_flow", sizeof(struct sw_flow), 0,
0, NULL);
if (flow_cache == NULL)
@@ -2016,7 +2038,7 @@ static bool ovs_sw_flow_mask_equal(const struct sw_flow_mask *a,
return (a->range.end == b->range.end)
&& (a->range.start == b->range.start)
- && (memcmp(a_, b_, ovs_sw_flow_mask_actual_size(a)) == 0);
+ && (memcmp(a_, b_, range_n_bytes(&a->range)) == 0);
}
struct sw_flow_mask *ovs_sw_flow_mask_find(const struct flow_table *tbl,
@@ -2053,5 +2075,5 @@ static void ovs_sw_flow_mask_set(struct sw_flow_mask *mask,
u8 *m = (u8 *)&mask->key + range->start;
mask->range = *range;
- memset(m, val, ovs_sw_flow_mask_size_roundup(mask));
+ memset(m, val, range_n_bytes(range));
}
diff --git a/net/openvswitch/flow.h b/net/openvswitch/flow.h
index e793051..b65f885 100644
--- a/net/openvswitch/flow.h
+++ b/net/openvswitch/flow.h
@@ -125,7 +125,7 @@ struct sw_flow_key {
} nd;
} ipv6;
};
-};
+} __aligned(__alignof__(long));
struct sw_flow {
struct rcu_head rcu;
@@ -149,11 +149,6 @@ struct sw_flow_key_range {
size_t end;
};
-static inline u16 ovs_sw_flow_key_range_actual_size(const struct sw_flow_key_range *range)
-{
- return range->end - range->start;
-}
-
struct sw_flow_match {
struct sw_flow_key *key;
struct sw_flow_key_range range;
@@ -253,18 +248,6 @@ struct sw_flow_mask {
struct sw_flow_key key;
};
-static inline u16
-ovs_sw_flow_mask_actual_size(const struct sw_flow_mask *mask)
-{
- return ovs_sw_flow_key_range_actual_size(&mask->range);
-}
-
-static inline u16
-ovs_sw_flow_mask_size_roundup(const struct sw_flow_mask *mask)
-{
- return roundup(ovs_sw_flow_mask_actual_size(mask), sizeof(u32));
-}
-
struct sw_flow_mask *ovs_sw_flow_mask_alloc(void);
void ovs_sw_flow_mask_add_ref(struct sw_flow_mask *);
void ovs_sw_flow_mask_del_ref(struct sw_flow_mask *, bool deferred);
--
1.8.1.2
^ permalink raw reply related
* [PATCH net-next 10/11] openvswitch: Rename key_len to key_end
From: Jesse Gross @ 2013-08-27 20:20 UTC (permalink / raw)
To: David Miller; +Cc: netdev, dev, Andy Zhou, Jesse Gross
In-Reply-To: <1377634848-34327-1-git-send-email-jesse@nicira.com>
From: Andy Zhou <azhou@nicira.com>
Key_end is a better name describing the ending boundary than key_len.
Rename those variables to make it less confusing.
Signed-off-by: Andy Zhou <azhou@nicira.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>
---
net/openvswitch/flow.c | 31 ++++++++++++++++---------------
net/openvswitch/flow.h | 2 +-
2 files changed, 17 insertions(+), 16 deletions(-)
diff --git a/net/openvswitch/flow.c b/net/openvswitch/flow.c
index 2b47855..80bcb96 100644
--- a/net/openvswitch/flow.c
+++ b/net/openvswitch/flow.c
@@ -997,10 +997,11 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key)
return 0;
}
-static u32 ovs_flow_hash(const struct sw_flow_key *key, int key_start, int key_len)
+static u32 ovs_flow_hash(const struct sw_flow_key *key, int key_start,
+ int key_end)
{
return jhash2((u32 *)((u8 *)key + key_start),
- DIV_ROUND_UP(key_len - key_start, sizeof(u32)), 0);
+ DIV_ROUND_UP(key_end - key_start, sizeof(u32)), 0);
}
static int flow_key_start(const struct sw_flow_key *key)
@@ -1012,31 +1013,31 @@ static int flow_key_start(const struct sw_flow_key *key)
}
static bool __cmp_key(const struct sw_flow_key *key1,
- const struct sw_flow_key *key2, int key_start, int key_len)
+ const struct sw_flow_key *key2, int key_start, int key_end)
{
return !memcmp((u8 *)key1 + key_start,
- (u8 *)key2 + key_start, (key_len - key_start));
+ (u8 *)key2 + key_start, (key_end - key_start));
}
static bool __flow_cmp_key(const struct sw_flow *flow,
- const struct sw_flow_key *key, int key_start, int key_len)
+ const struct sw_flow_key *key, int key_start, int key_end)
{
- return __cmp_key(&flow->key, key, key_start, key_len);
+ return __cmp_key(&flow->key, key, key_start, key_end);
}
static bool __flow_cmp_unmasked_key(const struct sw_flow *flow,
- const struct sw_flow_key *key, int key_start, int key_len)
+ const struct sw_flow_key *key, int key_start, int key_end)
{
- return __cmp_key(&flow->unmasked_key, key, key_start, key_len);
+ return __cmp_key(&flow->unmasked_key, key, key_start, key_end);
}
bool ovs_flow_cmp_unmasked_key(const struct sw_flow *flow,
- const struct sw_flow_key *key, int key_len)
+ const struct sw_flow_key *key, int key_end)
{
int key_start;
key_start = flow_key_start(key);
- return __flow_cmp_unmasked_key(flow, key, key_start, key_len);
+ return __flow_cmp_unmasked_key(flow, key, key_start, key_end);
}
@@ -1044,11 +1045,11 @@ struct sw_flow *ovs_flow_lookup_unmasked_key(struct flow_table *table,
struct sw_flow_match *match)
{
struct sw_flow_key *unmasked = match->key;
- int key_len = match->range.end;
+ int key_end = match->range.end;
struct sw_flow *flow;
flow = ovs_flow_lookup(table, unmasked);
- if (flow && (!ovs_flow_cmp_unmasked_key(flow, unmasked, key_len)))
+ if (flow && (!ovs_flow_cmp_unmasked_key(flow, unmasked, key_end)))
flow = NULL;
return flow;
@@ -1061,16 +1062,16 @@ static struct sw_flow *ovs_masked_flow_lookup(struct flow_table *table,
struct sw_flow *flow;
struct hlist_head *head;
int key_start = mask->range.start;
- int key_len = mask->range.end;
+ int key_end = mask->range.end;
u32 hash;
struct sw_flow_key masked_key;
ovs_flow_key_mask(&masked_key, flow_key, mask);
- hash = ovs_flow_hash(&masked_key, key_start, key_len);
+ hash = ovs_flow_hash(&masked_key, key_start, key_end);
head = find_bucket(table, hash);
hlist_for_each_entry_rcu(flow, head, hash_node[table->node_ver]) {
if (flow->mask == mask &&
- __flow_cmp_key(flow, &masked_key, key_start, key_len))
+ __flow_cmp_key(flow, &masked_key, key_start, key_end))
return flow;
}
return NULL;
diff --git a/net/openvswitch/flow.h b/net/openvswitch/flow.h
index d08dcf7..e793051 100644
--- a/net/openvswitch/flow.h
+++ b/net/openvswitch/flow.h
@@ -243,7 +243,7 @@ int ovs_ipv4_tun_to_nlattr(struct sk_buff *skb,
const struct ovs_key_ipv4_tunnel *output);
bool ovs_flow_cmp_unmasked_key(const struct sw_flow *flow,
- const struct sw_flow_key *key, int key_len);
+ const struct sw_flow_key *key, int key_end);
struct sw_flow_mask {
int ref_count;
--
1.8.1.2
^ permalink raw reply related
* [PATCH net-next 09/11] openvswitch: Add SCTP support
From: Jesse Gross @ 2013-08-27 20:20 UTC (permalink / raw)
To: David Miller
Cc: netdev, dev, Joe Stringer, Ben Pfaff, Simon Horman, Jesse Gross
In-Reply-To: <1377634848-34327-1-git-send-email-jesse@nicira.com>
From: Joe Stringer <joe@wand.net.nz>
This patch adds support for rewriting SCTP src,dst ports similar to the
functionality already available for TCP/UDP.
Rewriting SCTP ports is expensive due to double-recalculation of the
SCTP checksums; this is performed to ensure that packets traversing OVS
with invalid checksums will continue to the destination with any
checksum corruption intact.
Reviewed-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Joe Stringer <joe@wand.net.nz>
Signed-off-by: Ben Pfaff <blp@nicira.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>
---
include/uapi/linux/openvswitch.h | 6 ++++
net/openvswitch/Kconfig | 1 +
net/openvswitch/actions.c | 39 ++++++++++++++++++++++++
net/openvswitch/datapath.c | 6 ++++
net/openvswitch/flow.c | 65 ++++++++++++++++++++++++++++++++++++++++
net/openvswitch/flow.h | 8 ++---
6 files changed, 121 insertions(+), 4 deletions(-)
diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswitch.h
index de1fa5d..a74d375 100644
--- a/include/uapi/linux/openvswitch.h
+++ b/include/uapi/linux/openvswitch.h
@@ -259,6 +259,7 @@ enum ovs_key_attr {
OVS_KEY_ATTR_ND, /* struct ovs_key_nd */
OVS_KEY_ATTR_SKB_MARK, /* u32 skb mark */
OVS_KEY_ATTR_TUNNEL, /* Nested set of ovs_tunnel attributes */
+ OVS_KEY_ATTR_SCTP, /* struct ovs_key_sctp */
#ifdef __KERNEL__
OVS_KEY_ATTR_IPV4_TUNNEL, /* struct ovs_key_ipv4_tunnel */
@@ -333,6 +334,11 @@ struct ovs_key_udp {
__be16 udp_dst;
};
+struct ovs_key_sctp {
+ __be16 sctp_src;
+ __be16 sctp_dst;
+};
+
struct ovs_key_icmp {
__u8 icmp_type;
__u8 icmp_code;
diff --git a/net/openvswitch/Kconfig b/net/openvswitch/Kconfig
index bed30e6..6ecf491 100644
--- a/net/openvswitch/Kconfig
+++ b/net/openvswitch/Kconfig
@@ -4,6 +4,7 @@
config OPENVSWITCH
tristate "Open vSwitch"
+ select LIBCRC32C
---help---
Open vSwitch is a multilayer Ethernet switch targeted at virtualized
environments. In addition to supporting a variety of features
diff --git a/net/openvswitch/actions.c b/net/openvswitch/actions.c
index 1f68022..65cfaa8 100644
--- a/net/openvswitch/actions.c
+++ b/net/openvswitch/actions.c
@@ -22,6 +22,7 @@
#include <linux/in.h>
#include <linux/ip.h>
#include <linux/openvswitch.h>
+#include <linux/sctp.h>
#include <linux/tcp.h>
#include <linux/udp.h>
#include <linux/in6.h>
@@ -31,6 +32,7 @@
#include <net/ipv6.h>
#include <net/checksum.h>
#include <net/dsfield.h>
+#include <net/sctp/checksum.h>
#include "datapath.h"
#include "vport.h"
@@ -352,6 +354,39 @@ static int set_tcp(struct sk_buff *skb, const struct ovs_key_tcp *tcp_port_key)
return 0;
}
+static int set_sctp(struct sk_buff *skb,
+ const struct ovs_key_sctp *sctp_port_key)
+{
+ struct sctphdr *sh;
+ int err;
+ unsigned int sctphoff = skb_transport_offset(skb);
+
+ err = make_writable(skb, sctphoff + sizeof(struct sctphdr));
+ if (unlikely(err))
+ return err;
+
+ sh = sctp_hdr(skb);
+ if (sctp_port_key->sctp_src != sh->source ||
+ sctp_port_key->sctp_dst != sh->dest) {
+ __le32 old_correct_csum, new_csum, old_csum;
+
+ old_csum = sh->checksum;
+ old_correct_csum = sctp_compute_cksum(skb, sctphoff);
+
+ sh->source = sctp_port_key->sctp_src;
+ sh->dest = sctp_port_key->sctp_dst;
+
+ new_csum = sctp_compute_cksum(skb, sctphoff);
+
+ /* Carry any checksum errors through. */
+ sh->checksum = old_csum ^ old_correct_csum ^ new_csum;
+
+ skb->rxhash = 0;
+ }
+
+ return 0;
+}
+
static int do_output(struct datapath *dp, struct sk_buff *skb, int out_port)
{
struct vport *vport;
@@ -461,6 +496,10 @@ static int execute_set_action(struct sk_buff *skb,
case OVS_KEY_ATTR_UDP:
err = set_udp(skb, nla_data(nested_attr));
break;
+
+ case OVS_KEY_ATTR_SCTP:
+ err = set_sctp(skb, nla_data(nested_attr));
+ break;
}
return err;
diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index d29cd9a..2aa13bd 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -712,6 +712,12 @@ static int validate_set(const struct nlattr *a,
return validate_tp_port(flow_key);
+ case OVS_KEY_ATTR_SCTP:
+ if (flow_key->ip.proto != IPPROTO_SCTP)
+ return -EINVAL;
+
+ return validate_tp_port(flow_key);
+
default:
return -EINVAL;
}
diff --git a/net/openvswitch/flow.c b/net/openvswitch/flow.c
index 1fceb96..2b47855 100644
--- a/net/openvswitch/flow.c
+++ b/net/openvswitch/flow.c
@@ -34,6 +34,7 @@
#include <linux/if_arp.h>
#include <linux/ip.h>
#include <linux/ipv6.h>
+#include <linux/sctp.h>
#include <linux/tcp.h>
#include <linux/udp.h>
#include <linux/icmp.h>
@@ -129,6 +130,7 @@ static bool ovs_match_validate(const struct sw_flow_match *match,
| (1 << OVS_KEY_ATTR_IPV6)
| (1 << OVS_KEY_ATTR_TCP)
| (1 << OVS_KEY_ATTR_UDP)
+ | (1 << OVS_KEY_ATTR_SCTP)
| (1 << OVS_KEY_ATTR_ICMP)
| (1 << OVS_KEY_ATTR_ICMPV6)
| (1 << OVS_KEY_ATTR_ARP)
@@ -159,6 +161,12 @@ static bool ovs_match_validate(const struct sw_flow_match *match,
mask_allowed |= 1 << OVS_KEY_ATTR_UDP;
}
+ if (match->key->ip.proto == IPPROTO_SCTP) {
+ key_expected |= 1 << OVS_KEY_ATTR_SCTP;
+ if (match->mask && (match->mask->key.ip.proto == 0xff))
+ mask_allowed |= 1 << OVS_KEY_ATTR_SCTP;
+ }
+
if (match->key->ip.proto == IPPROTO_TCP) {
key_expected |= 1 << OVS_KEY_ATTR_TCP;
if (match->mask && (match->mask->key.ip.proto == 0xff))
@@ -185,6 +193,12 @@ static bool ovs_match_validate(const struct sw_flow_match *match,
mask_allowed |= 1 << OVS_KEY_ATTR_UDP;
}
+ if (match->key->ip.proto == IPPROTO_SCTP) {
+ key_expected |= 1 << OVS_KEY_ATTR_SCTP;
+ if (match->mask && (match->mask->key.ip.proto == 0xff))
+ mask_allowed |= 1 << OVS_KEY_ATTR_SCTP;
+ }
+
if (match->key->ip.proto == IPPROTO_TCP) {
key_expected |= 1 << OVS_KEY_ATTR_TCP;
if (match->mask && (match->mask->key.ip.proto == 0xff))
@@ -280,6 +294,12 @@ static bool udphdr_ok(struct sk_buff *skb)
sizeof(struct udphdr));
}
+static bool sctphdr_ok(struct sk_buff *skb)
+{
+ return pskb_may_pull(skb, skb_transport_offset(skb) +
+ sizeof(struct sctphdr));
+}
+
static bool icmphdr_ok(struct sk_buff *skb)
{
return pskb_may_pull(skb, skb_transport_offset(skb) +
@@ -891,6 +911,12 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key)
key->ipv4.tp.src = udp->source;
key->ipv4.tp.dst = udp->dest;
}
+ } else if (key->ip.proto == IPPROTO_SCTP) {
+ if (sctphdr_ok(skb)) {
+ struct sctphdr *sctp = sctp_hdr(skb);
+ key->ipv4.tp.src = sctp->source;
+ key->ipv4.tp.dst = sctp->dest;
+ }
} else if (key->ip.proto == IPPROTO_ICMP) {
if (icmphdr_ok(skb)) {
struct icmphdr *icmp = icmp_hdr(skb);
@@ -953,6 +979,12 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key)
key->ipv6.tp.src = udp->source;
key->ipv6.tp.dst = udp->dest;
}
+ } else if (key->ip.proto == NEXTHDR_SCTP) {
+ if (sctphdr_ok(skb)) {
+ struct sctphdr *sctp = sctp_hdr(skb);
+ key->ipv6.tp.src = sctp->source;
+ key->ipv6.tp.dst = sctp->dest;
+ }
} else if (key->ip.proto == NEXTHDR_ICMP) {
if (icmp6hdr_ok(skb)) {
error = parse_icmpv6(skb, key, nh_len);
@@ -1087,6 +1119,7 @@ const int ovs_key_lens[OVS_KEY_ATTR_MAX + 1] = {
[OVS_KEY_ATTR_IPV6] = sizeof(struct ovs_key_ipv6),
[OVS_KEY_ATTR_TCP] = sizeof(struct ovs_key_tcp),
[OVS_KEY_ATTR_UDP] = sizeof(struct ovs_key_udp),
+ [OVS_KEY_ATTR_SCTP] = sizeof(struct ovs_key_sctp),
[OVS_KEY_ATTR_ICMP] = sizeof(struct ovs_key_icmp),
[OVS_KEY_ATTR_ICMPV6] = sizeof(struct ovs_key_icmpv6),
[OVS_KEY_ATTR_ARP] = sizeof(struct ovs_key_arp),
@@ -1500,6 +1533,24 @@ static int ovs_key_from_nlattrs(struct sw_flow_match *match, u64 attrs,
attrs &= ~(1 << OVS_KEY_ATTR_UDP);
}
+ if (attrs & (1 << OVS_KEY_ATTR_SCTP)) {
+ const struct ovs_key_sctp *sctp_key;
+
+ sctp_key = nla_data(a[OVS_KEY_ATTR_SCTP]);
+ if (orig_attrs & (1 << OVS_KEY_ATTR_IPV4)) {
+ SW_FLOW_KEY_PUT(match, ipv4.tp.src,
+ sctp_key->sctp_src, is_mask);
+ SW_FLOW_KEY_PUT(match, ipv4.tp.dst,
+ sctp_key->sctp_dst, is_mask);
+ } else {
+ SW_FLOW_KEY_PUT(match, ipv6.tp.src,
+ sctp_key->sctp_src, is_mask);
+ SW_FLOW_KEY_PUT(match, ipv6.tp.dst,
+ sctp_key->sctp_dst, is_mask);
+ }
+ attrs &= ~(1 << OVS_KEY_ATTR_SCTP);
+ }
+
if (attrs & (1 << OVS_KEY_ATTR_ICMP)) {
const struct ovs_key_icmp *icmp_key;
@@ -1843,6 +1894,20 @@ int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey,
udp_key->udp_src = output->ipv6.tp.src;
udp_key->udp_dst = output->ipv6.tp.dst;
}
+ } else if (swkey->ip.proto == IPPROTO_SCTP) {
+ struct ovs_key_sctp *sctp_key;
+
+ nla = nla_reserve(skb, OVS_KEY_ATTR_SCTP, sizeof(*sctp_key));
+ if (!nla)
+ goto nla_put_failure;
+ sctp_key = nla_data(nla);
+ if (swkey->eth.type == htons(ETH_P_IP)) {
+ sctp_key->sctp_src = swkey->ipv4.tp.src;
+ sctp_key->sctp_dst = swkey->ipv4.tp.dst;
+ } else if (swkey->eth.type == htons(ETH_P_IPV6)) {
+ sctp_key->sctp_src = swkey->ipv6.tp.src;
+ sctp_key->sctp_dst = swkey->ipv6.tp.dst;
+ }
} else if (swkey->eth.type == htons(ETH_P_IP) &&
swkey->ip.proto == IPPROTO_ICMP) {
struct ovs_key_icmp *icmp_key;
diff --git a/net/openvswitch/flow.h b/net/openvswitch/flow.h
index 9674e45..d08dcf7 100644
--- a/net/openvswitch/flow.h
+++ b/net/openvswitch/flow.h
@@ -99,8 +99,8 @@ struct sw_flow_key {
} addr;
union {
struct {
- __be16 src; /* TCP/UDP source port. */
- __be16 dst; /* TCP/UDP destination port. */
+ __be16 src; /* TCP/UDP/SCTP source port. */
+ __be16 dst; /* TCP/UDP/SCTP destination port. */
} tp;
struct {
u8 sha[ETH_ALEN]; /* ARP source hardware address. */
@@ -115,8 +115,8 @@ struct sw_flow_key {
} addr;
__be32 label; /* IPv6 flow label. */
struct {
- __be16 src; /* TCP/UDP source port. */
- __be16 dst; /* TCP/UDP destination port. */
+ __be16 src; /* TCP/UDP/SCTP source port. */
+ __be16 dst; /* TCP/UDP/SCTP destination port. */
} tp;
struct {
struct in6_addr target; /* ND target address. */
--
1.8.1.2
^ permalink raw reply related
* [PATCH net-next 08/11] net: Add NEXTHDR_SCTP to ipv6.h
From: Jesse Gross @ 2013-08-27 20:20 UTC (permalink / raw)
To: David Miller; +Cc: netdev, dev, Joe Stringer, Jesse Gross
In-Reply-To: <1377634848-34327-1-git-send-email-jesse@nicira.com>
From: Joe Stringer <joe@wand.net.nz>
Signed-off-by: Joe Stringer <joe@wand.net.nz>
Signed-off-by: Jesse Gross <jesse@nicira.com>
---
include/net/ipv6.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/include/net/ipv6.h b/include/net/ipv6.h
index 5fe5649..7bdff04 100644
--- a/include/net/ipv6.h
+++ b/include/net/ipv6.h
@@ -41,6 +41,7 @@
#define NEXTHDR_ICMP 58 /* ICMP for IPv6. */
#define NEXTHDR_NONE 59 /* No next header */
#define NEXTHDR_DEST 60 /* Destination options header. */
+#define NEXTHDR_SCTP 132 /* SCTP message. */
#define NEXTHDR_MOBILITY 135 /* Mobility header. */
#define NEXTHDR_MAX 255
--
1.8.1.2
^ permalink raw reply related
* [PATCH net-next 07/11] openvswitch: Mega flow implementation
From: Jesse Gross @ 2013-08-27 20:20 UTC (permalink / raw)
To: David Miller; +Cc: netdev, dev, Andy Zhou, Pravin B Shelar, Jesse Gross
In-Reply-To: <1377634848-34327-1-git-send-email-jesse@nicira.com>
From: Andy Zhou <azhou@nicira.com>
Add wildcarded flow support in kernel datapath.
Wildcarded flow can improve OVS flow set up performance by avoid sending
matching new flows to the user space program. The exact performance boost
will largely dependent on wildcarded flow hit rate.
In case all new flows hits wildcard flows, the flow set up rate is
within 5% of that of linux bridge module.
Pravin has made significant contributions to this patch. Including API
clean ups and bug fixes.
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: Andy Zhou <azhou@nicira.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>
---
Documentation/networking/openvswitch.txt | 40 +
include/uapi/linux/openvswitch.h | 9 +-
net/openvswitch/actions.c | 6 +-
net/openvswitch/datapath.c | 140 ++-
net/openvswitch/datapath.h | 6 +
net/openvswitch/flow.c | 1387 ++++++++++++++++++++----------
net/openvswitch/flow.h | 96 ++-
7 files changed, 1171 insertions(+), 513 deletions(-)
diff --git a/Documentation/networking/openvswitch.txt b/Documentation/networking/openvswitch.txt
index 8fa2dd1..37c20ee 100644
--- a/Documentation/networking/openvswitch.txt
+++ b/Documentation/networking/openvswitch.txt
@@ -91,6 +91,46 @@ Often we ellipsize arguments not important to the discussion, e.g.:
in_port(1), eth(...), eth_type(0x0800), ipv4(...), tcp(...)
+Wildcarded flow key format
+--------------------------
+
+A wildcarded flow is described with two sequences of Netlink attributes
+passed over the Netlink socket. A flow key, exactly as described above, and an
+optional corresponding flow mask.
+
+A wildcarded flow can represent a group of exact match flows. Each '1' bit
+in the mask specifies a exact match with the corresponding bit in the flow key.
+A '0' bit specifies a don't care bit, which will match either a '1' or '0' bit
+of a incoming packet. Using wildcarded flow can improve the flow set up rate
+by reduce the number of new flows need to be processed by the user space program.
+
+Support for the mask Netlink attribute is optional for both the kernel and user
+space program. The kernel can ignore the mask attribute, installing an exact
+match flow, or reduce the number of don't care bits in the kernel to less than
+what was specified by the user space program. In this case, variations in bits
+that the kernel does not implement will simply result in additional flow setups.
+The kernel module will also work with user space programs that neither support
+nor supply flow mask attributes.
+
+Since the kernel may ignore or modify wildcard bits, it can be difficult for
+the userspace program to know exactly what matches are installed. There are
+two possible approaches: reactively install flows as they miss the kernel
+flow table (and therefore not attempt to determine wildcard changes at all)
+or use the kernel's response messages to determine the installed wildcards.
+
+When interacting with userspace, the kernel should maintain the match portion
+of the key exactly as originally installed. This will provides a handle to
+identify the flow for all future operations. However, when reporting the
+mask of an installed flow, the mask should include any restrictions imposed
+by the kernel.
+
+The behavior when using overlapping wildcarded flows is undefined. It is the
+responsibility of the user space program to ensure that any incoming packet
+can match at most one flow, wildcarded or not. The current implementation
+performs best-effort detection of overlapping wildcarded flows and may reject
+some but not all of them. However, this behavior may change in future versions.
+
+
Basic rule for evolving flow keys
---------------------------------
diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswitch.h
index 52490b0..de1fa5d 100644
--- a/include/uapi/linux/openvswitch.h
+++ b/include/uapi/linux/openvswitch.h
@@ -1,6 +1,6 @@
/*
- * Copyright (c) 2007-2011 Nicira Networks.
+ * Copyright (c) 2007-2013 Nicira, Inc.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of version 2 of the GNU General Public
@@ -379,6 +379,12 @@ struct ovs_key_nd {
* @OVS_FLOW_ATTR_CLEAR: If present in a %OVS_FLOW_CMD_SET request, clears the
* last-used time, accumulated TCP flags, and statistics for this flow.
* Otherwise ignored in requests. Never present in notifications.
+ * @OVS_FLOW_ATTR_MASK: Nested %OVS_KEY_ATTR_* attributes specifying the
+ * mask bits for wildcarded flow match. Mask bit value '1' specifies exact
+ * match with corresponding flow key bit, while mask bit value '0' specifies
+ * a wildcarded match. Omitting attribute is treated as wildcarding all
+ * corresponding fields. Optional for all requests. If not present,
+ * all flow key bits are exact match bits.
*
* These attributes follow the &struct ovs_header within the Generic Netlink
* payload for %OVS_FLOW_* commands.
@@ -391,6 +397,7 @@ enum ovs_flow_attr {
OVS_FLOW_ATTR_TCP_FLAGS, /* 8-bit OR'd TCP flags. */
OVS_FLOW_ATTR_USED, /* u64 msecs last used in monotonic time. */
OVS_FLOW_ATTR_CLEAR, /* Flag to clear stats, tcp_flags, used. */
+ OVS_FLOW_ATTR_MASK, /* Sequence of OVS_KEY_ATTR_* attributes. */
__OVS_FLOW_ATTR_MAX
};
diff --git a/net/openvswitch/actions.c b/net/openvswitch/actions.c
index ab101f7..1f68022 100644
--- a/net/openvswitch/actions.c
+++ b/net/openvswitch/actions.c
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2007-2012 Nicira, Inc.
+ * Copyright (c) 2007-2013 Nicira, Inc.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of version 2 of the GNU General Public
@@ -376,8 +376,10 @@ static int output_userspace(struct datapath *dp, struct sk_buff *skb,
const struct nlattr *a;
int rem;
+ BUG_ON(!OVS_CB(skb)->pkt_key);
+
upcall.cmd = OVS_PACKET_CMD_ACTION;
- upcall.key = &OVS_CB(skb)->flow->key;
+ upcall.key = OVS_CB(skb)->pkt_key;
upcall.userdata = NULL;
upcall.portid = 0;
diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index 9d97ef3..d29cd9a 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2007-2012 Nicira, Inc.
+ * Copyright (c) 2007-2013 Nicira, Inc.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of version 2 of the GNU General Public
@@ -165,7 +165,7 @@ static void destroy_dp_rcu(struct rcu_head *rcu)
{
struct datapath *dp = container_of(rcu, struct datapath, rcu);
- ovs_flow_tbl_destroy((__force struct flow_table *)dp->table);
+ ovs_flow_tbl_destroy((__force struct flow_table *)dp->table, false);
free_percpu(dp->stats_percpu);
release_net(ovs_dp_get_net(dp));
kfree(dp->ports);
@@ -226,19 +226,18 @@ void ovs_dp_process_received_packet(struct vport *p, struct sk_buff *skb)
struct sw_flow_key key;
u64 *stats_counter;
int error;
- int key_len;
stats = this_cpu_ptr(dp->stats_percpu);
/* Extract flow from 'skb' into 'key'. */
- error = ovs_flow_extract(skb, p->port_no, &key, &key_len);
+ error = ovs_flow_extract(skb, p->port_no, &key);
if (unlikely(error)) {
kfree_skb(skb);
return;
}
/* Look up flow. */
- flow = ovs_flow_tbl_lookup(rcu_dereference(dp->table), &key, key_len);
+ flow = ovs_flow_lookup(rcu_dereference(dp->table), &key);
if (unlikely(!flow)) {
struct dp_upcall_info upcall;
@@ -253,6 +252,7 @@ void ovs_dp_process_received_packet(struct vport *p, struct sk_buff *skb)
}
OVS_CB(skb)->flow = flow;
+ OVS_CB(skb)->pkt_key = &key;
stats_counter = &stats->n_hit;
ovs_flow_used(OVS_CB(skb)->flow, skb);
@@ -435,7 +435,7 @@ static int queue_userspace_packet(struct net *net, int dp_ifindex,
upcall->dp_ifindex = dp_ifindex;
nla = nla_nest_start(user_skb, OVS_PACKET_ATTR_KEY);
- ovs_flow_to_nlattrs(upcall_info->key, user_skb);
+ ovs_flow_to_nlattrs(upcall_info->key, upcall_info->key, user_skb);
nla_nest_end(user_skb, nla);
if (upcall_info->userdata)
@@ -468,7 +468,7 @@ static int flush_flows(struct datapath *dp)
rcu_assign_pointer(dp->table, new_table);
- ovs_flow_tbl_deferred_destroy(old_table);
+ ovs_flow_tbl_destroy(old_table, true);
return 0;
}
@@ -611,10 +611,12 @@ static int validate_tp_port(const struct sw_flow_key *flow_key)
static int validate_and_copy_set_tun(const struct nlattr *attr,
struct sw_flow_actions **sfa)
{
- struct ovs_key_ipv4_tunnel tun_key;
+ struct sw_flow_match match;
+ struct sw_flow_key key;
int err, start;
- err = ovs_ipv4_tun_from_nlattr(nla_data(attr), &tun_key);
+ ovs_match_init(&match, &key, NULL);
+ err = ovs_ipv4_tun_from_nlattr(nla_data(attr), &match, false);
if (err)
return err;
@@ -622,7 +624,8 @@ static int validate_and_copy_set_tun(const struct nlattr *attr,
if (start < 0)
return start;
- err = add_action(sfa, OVS_KEY_ATTR_IPV4_TUNNEL, &tun_key, sizeof(tun_key));
+ err = add_action(sfa, OVS_KEY_ATTR_IPV4_TUNNEL, &match.key->tun_key,
+ sizeof(match.key->tun_key));
add_nested_action_end(*sfa, start);
return err;
@@ -857,7 +860,6 @@ static int ovs_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info)
struct ethhdr *eth;
int len;
int err;
- int key_len;
err = -EINVAL;
if (!a[OVS_PACKET_ATTR_PACKET] || !a[OVS_PACKET_ATTR_KEY] ||
@@ -890,11 +892,11 @@ static int ovs_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info)
if (IS_ERR(flow))
goto err_kfree_skb;
- err = ovs_flow_extract(packet, -1, &flow->key, &key_len);
+ err = ovs_flow_extract(packet, -1, &flow->key);
if (err)
goto err_flow_free;
- err = ovs_flow_metadata_from_nlattrs(flow, key_len, a[OVS_PACKET_ATTR_KEY]);
+ err = ovs_flow_metadata_from_nlattrs(flow, a[OVS_PACKET_ATTR_KEY]);
if (err)
goto err_flow_free;
acts = ovs_flow_actions_alloc(nla_len(a[OVS_PACKET_ATTR_ACTIONS]));
@@ -908,6 +910,7 @@ static int ovs_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info)
goto err_flow_free;
OVS_CB(packet)->flow = flow;
+ OVS_CB(packet)->pkt_key = &flow->key;
packet->priority = flow->key.phy.priority;
packet->mark = flow->key.phy.skb_mark;
@@ -922,13 +925,13 @@ static int ovs_packet_cmd_execute(struct sk_buff *skb, struct genl_info *info)
local_bh_enable();
rcu_read_unlock();
- ovs_flow_free(flow);
+ ovs_flow_free(flow, false);
return err;
err_unlock:
rcu_read_unlock();
err_flow_free:
- ovs_flow_free(flow);
+ ovs_flow_free(flow, false);
err_kfree_skb:
kfree_skb(packet);
err:
@@ -1045,7 +1048,8 @@ static int set_action_to_attr(const struct nlattr *a, struct sk_buff *skb)
if (!start)
return -EMSGSIZE;
- err = ovs_ipv4_tun_to_nlattr(skb, nla_data(ovs_key));
+ err = ovs_ipv4_tun_to_nlattr(skb, nla_data(ovs_key),
+ nla_data(ovs_key));
if (err)
return err;
nla_nest_end(skb, start);
@@ -1093,6 +1097,7 @@ static size_t ovs_flow_cmd_msg_size(const struct sw_flow_actions *acts)
{
return NLMSG_ALIGN(sizeof(struct ovs_header))
+ nla_total_size(key_attr_size()) /* OVS_FLOW_ATTR_KEY */
+ + nla_total_size(key_attr_size()) /* OVS_FLOW_ATTR_MASK */
+ nla_total_size(sizeof(struct ovs_flow_stats)) /* OVS_FLOW_ATTR_STATS */
+ nla_total_size(1) /* OVS_FLOW_ATTR_TCP_FLAGS */
+ nla_total_size(8) /* OVS_FLOW_ATTR_USED */
@@ -1119,12 +1124,25 @@ static int ovs_flow_cmd_fill_info(struct sw_flow *flow, struct datapath *dp,
ovs_header->dp_ifindex = get_dpifindex(dp);
+ /* Fill flow key. */
nla = nla_nest_start(skb, OVS_FLOW_ATTR_KEY);
if (!nla)
goto nla_put_failure;
- err = ovs_flow_to_nlattrs(&flow->key, skb);
+
+ err = ovs_flow_to_nlattrs(&flow->unmasked_key,
+ &flow->unmasked_key, skb);
+ if (err)
+ goto error;
+ nla_nest_end(skb, nla);
+
+ nla = nla_nest_start(skb, OVS_FLOW_ATTR_MASK);
+ if (!nla)
+ goto nla_put_failure;
+
+ err = ovs_flow_to_nlattrs(&flow->key, &flow->mask->key, skb);
if (err)
goto error;
+
nla_nest_end(skb, nla);
spin_lock_bh(&flow->lock);
@@ -1214,20 +1232,24 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
{
struct nlattr **a = info->attrs;
struct ovs_header *ovs_header = info->userhdr;
- struct sw_flow_key key;
- struct sw_flow *flow;
+ struct sw_flow_key key, masked_key;
+ struct sw_flow *flow = NULL;
+ struct sw_flow_mask mask;
struct sk_buff *reply;
struct datapath *dp;
struct flow_table *table;
struct sw_flow_actions *acts = NULL;
+ struct sw_flow_match match;
int error;
- int key_len;
/* Extract key. */
error = -EINVAL;
if (!a[OVS_FLOW_ATTR_KEY])
goto error;
- error = ovs_flow_from_nlattrs(&key, &key_len, a[OVS_FLOW_ATTR_KEY]);
+
+ ovs_match_init(&match, &key, &mask);
+ error = ovs_match_from_nlattrs(&match,
+ a[OVS_FLOW_ATTR_KEY], a[OVS_FLOW_ATTR_MASK]);
if (error)
goto error;
@@ -1238,9 +1260,13 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
if (IS_ERR(acts))
goto error;
- error = validate_and_copy_actions(a[OVS_FLOW_ATTR_ACTIONS], &key, 0, &acts);
- if (error)
+ ovs_flow_key_mask(&masked_key, &key, &mask);
+ error = validate_and_copy_actions(a[OVS_FLOW_ATTR_ACTIONS],
+ &masked_key, 0, &acts);
+ if (error) {
+ OVS_NLERR("Flow actions may not be safe on all matching packets.\n");
goto err_kfree;
+ }
} else if (info->genlhdr->cmd == OVS_FLOW_CMD_NEW) {
error = -EINVAL;
goto error;
@@ -1253,8 +1279,11 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
goto err_unlock_ovs;
table = ovsl_dereference(dp->table);
- flow = ovs_flow_tbl_lookup(table, &key, key_len);
+
+ /* Check if this is a duplicate flow */
+ flow = ovs_flow_lookup(table, &key);
if (!flow) {
+ struct sw_flow_mask *mask_p;
/* Bail out if we're not allowed to create a new flow. */
error = -ENOENT;
if (info->genlhdr->cmd == OVS_FLOW_CMD_SET)
@@ -1267,7 +1296,7 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
new_table = ovs_flow_tbl_expand(table);
if (!IS_ERR(new_table)) {
rcu_assign_pointer(dp->table, new_table);
- ovs_flow_tbl_deferred_destroy(table);
+ ovs_flow_tbl_destroy(table, true);
table = ovsl_dereference(dp->table);
}
}
@@ -1280,14 +1309,30 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
}
clear_stats(flow);
+ flow->key = masked_key;
+ flow->unmasked_key = key;
+
+ /* Make sure mask is unique in the system */
+ mask_p = ovs_sw_flow_mask_find(table, &mask);
+ if (!mask_p) {
+ /* Allocate a new mask if none exsits. */
+ mask_p = ovs_sw_flow_mask_alloc();
+ if (!mask_p)
+ goto err_flow_free;
+ mask_p->key = mask.key;
+ mask_p->range = mask.range;
+ ovs_sw_flow_mask_insert(table, mask_p);
+ }
+
+ ovs_sw_flow_mask_add_ref(mask_p);
+ flow->mask = mask_p;
rcu_assign_pointer(flow->sf_acts, acts);
/* Put flow in bucket. */
- ovs_flow_tbl_insert(table, flow, &key, key_len);
+ ovs_flow_insert(table, flow);
reply = ovs_flow_cmd_build_info(flow, dp, info->snd_portid,
- info->snd_seq,
- OVS_FLOW_CMD_NEW);
+ info->snd_seq, OVS_FLOW_CMD_NEW);
} else {
/* We found a matching flow. */
struct sw_flow_actions *old_acts;
@@ -1303,6 +1348,13 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
info->nlhdr->nlmsg_flags & (NLM_F_CREATE | NLM_F_EXCL))
goto err_unlock_ovs;
+ /* The unmasked key has to be the same for flow updates. */
+ error = -EINVAL;
+ if (!ovs_flow_cmp_unmasked_key(flow, &key, match.range.end)) {
+ OVS_NLERR("Flow modification message rejected, unmasked key does not match.\n");
+ goto err_unlock_ovs;
+ }
+
/* Update actions. */
old_acts = ovsl_dereference(flow->sf_acts);
rcu_assign_pointer(flow->sf_acts, acts);
@@ -1327,6 +1379,8 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
ovs_dp_flow_multicast_group.id, PTR_ERR(reply));
return 0;
+err_flow_free:
+ ovs_flow_free(flow, false);
err_unlock_ovs:
ovs_unlock();
err_kfree:
@@ -1344,12 +1398,16 @@ static int ovs_flow_cmd_get(struct sk_buff *skb, struct genl_info *info)
struct sw_flow *flow;
struct datapath *dp;
struct flow_table *table;
+ struct sw_flow_match match;
int err;
- int key_len;
- if (!a[OVS_FLOW_ATTR_KEY])
+ if (!a[OVS_FLOW_ATTR_KEY]) {
+ OVS_NLERR("Flow get message rejected, Key attribute missing.\n");
return -EINVAL;
- err = ovs_flow_from_nlattrs(&key, &key_len, a[OVS_FLOW_ATTR_KEY]);
+ }
+
+ ovs_match_init(&match, &key, NULL);
+ err = ovs_match_from_nlattrs(&match, a[OVS_FLOW_ATTR_KEY], NULL);
if (err)
return err;
@@ -1361,7 +1419,7 @@ static int ovs_flow_cmd_get(struct sk_buff *skb, struct genl_info *info)
}
table = ovsl_dereference(dp->table);
- flow = ovs_flow_tbl_lookup(table, &key, key_len);
+ flow = ovs_flow_lookup_unmasked_key(table, &match);
if (!flow) {
err = -ENOENT;
goto unlock;
@@ -1390,8 +1448,8 @@ static int ovs_flow_cmd_del(struct sk_buff *skb, struct genl_info *info)
struct sw_flow *flow;
struct datapath *dp;
struct flow_table *table;
+ struct sw_flow_match match;
int err;
- int key_len;
ovs_lock();
dp = get_dp(sock_net(skb->sk), ovs_header->dp_ifindex);
@@ -1404,12 +1462,14 @@ static int ovs_flow_cmd_del(struct sk_buff *skb, struct genl_info *info)
err = flush_flows(dp);
goto unlock;
}
- err = ovs_flow_from_nlattrs(&key, &key_len, a[OVS_FLOW_ATTR_KEY]);
+
+ ovs_match_init(&match, &key, NULL);
+ err = ovs_match_from_nlattrs(&match, a[OVS_FLOW_ATTR_KEY], NULL);
if (err)
goto unlock;
table = ovsl_dereference(dp->table);
- flow = ovs_flow_tbl_lookup(table, &key, key_len);
+ flow = ovs_flow_lookup_unmasked_key(table, &match);
if (!flow) {
err = -ENOENT;
goto unlock;
@@ -1421,13 +1481,13 @@ static int ovs_flow_cmd_del(struct sk_buff *skb, struct genl_info *info)
goto unlock;
}
- ovs_flow_tbl_remove(table, flow);
+ ovs_flow_remove(table, flow);
err = ovs_flow_cmd_fill_info(flow, dp, reply, info->snd_portid,
info->snd_seq, 0, OVS_FLOW_CMD_DEL);
BUG_ON(err < 0);
- ovs_flow_deferred_free(flow);
+ ovs_flow_free(flow, true);
ovs_unlock();
ovs_notify(reply, info, &ovs_dp_flow_multicast_group);
@@ -1457,7 +1517,7 @@ static int ovs_flow_cmd_dump(struct sk_buff *skb, struct netlink_callback *cb)
bucket = cb->args[0];
obj = cb->args[1];
- flow = ovs_flow_tbl_next(table, &bucket, &obj);
+ flow = ovs_flow_dump_next(table, &bucket, &obj);
if (!flow)
break;
@@ -1680,7 +1740,7 @@ err_destroy_ports_array:
err_destroy_percpu:
free_percpu(dp->stats_percpu);
err_destroy_table:
- ovs_flow_tbl_destroy(ovsl_dereference(dp->table));
+ ovs_flow_tbl_destroy(ovsl_dereference(dp->table), false);
err_free_dp:
release_net(ovs_dp_get_net(dp));
kfree(dp);
@@ -2287,7 +2347,7 @@ static void rehash_flow_table(struct work_struct *work)
new_table = ovs_flow_tbl_rehash(old_table);
if (!IS_ERR(new_table)) {
rcu_assign_pointer(dp->table, new_table);
- ovs_flow_tbl_deferred_destroy(old_table);
+ ovs_flow_tbl_destroy(old_table, true);
}
}
}
diff --git a/net/openvswitch/datapath.h b/net/openvswitch/datapath.h
index a914864..4d109c1 100644
--- a/net/openvswitch/datapath.h
+++ b/net/openvswitch/datapath.h
@@ -88,11 +88,13 @@ struct datapath {
/**
* struct ovs_skb_cb - OVS data in skb CB
* @flow: The flow associated with this packet. May be %NULL if no flow.
+ * @pkt_key: The flow information extracted from the packet. Must be nonnull.
* @tun_key: Key for the tunnel that encapsulated this packet. NULL if the
* packet is not being tunneled.
*/
struct ovs_skb_cb {
struct sw_flow *flow;
+ struct sw_flow_key *pkt_key;
struct ovs_key_ipv4_tunnel *tun_key;
};
#define OVS_CB(skb) ((struct ovs_skb_cb *)(skb)->cb)
@@ -183,4 +185,8 @@ struct sk_buff *ovs_vport_cmd_build_info(struct vport *, u32 pid, u32 seq,
int ovs_execute_actions(struct datapath *dp, struct sk_buff *skb);
void ovs_dp_notify_wq(struct work_struct *work);
+
+#define OVS_NLERR(fmt, ...) \
+ pr_info_once("netlink: " fmt, ##__VA_ARGS__)
+
#endif /* datapath.h */
diff --git a/net/openvswitch/flow.c b/net/openvswitch/flow.c
index fca2825..1fceb96 100644
--- a/net/openvswitch/flow.c
+++ b/net/openvswitch/flow.c
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2007-2011 Nicira, Inc.
+ * Copyright (c) 2007-2013 Nicira, Inc.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of version 2 of the GNU General Public
@@ -46,6 +46,184 @@
static struct kmem_cache *flow_cache;
+static void ovs_sw_flow_mask_set(struct sw_flow_mask *mask,
+ struct sw_flow_key_range *range, u8 val);
+
+static void update_range__(struct sw_flow_match *match,
+ size_t offset, size_t size, bool is_mask)
+{
+ struct sw_flow_key_range *range = NULL;
+ size_t start = offset;
+ size_t end = offset + size;
+
+ if (!is_mask)
+ range = &match->range;
+ else if (match->mask)
+ range = &match->mask->range;
+
+ if (!range)
+ return;
+
+ if (range->start == range->end) {
+ range->start = start;
+ range->end = end;
+ return;
+ }
+
+ if (range->start > start)
+ range->start = start;
+
+ if (range->end < end)
+ range->end = end;
+}
+
+#define SW_FLOW_KEY_PUT(match, field, value, is_mask) \
+ do { \
+ update_range__(match, offsetof(struct sw_flow_key, field), \
+ sizeof((match)->key->field), is_mask); \
+ if (is_mask) { \
+ if ((match)->mask) \
+ (match)->mask->key.field = value; \
+ } else { \
+ (match)->key->field = value; \
+ } \
+ } while (0)
+
+#define SW_FLOW_KEY_MEMCPY(match, field, value_p, len, is_mask) \
+ do { \
+ update_range__(match, offsetof(struct sw_flow_key, field), \
+ len, is_mask); \
+ if (is_mask) { \
+ if ((match)->mask) \
+ memcpy(&(match)->mask->key.field, value_p, len);\
+ } else { \
+ memcpy(&(match)->key->field, value_p, len); \
+ } \
+ } while (0)
+
+void ovs_match_init(struct sw_flow_match *match,
+ struct sw_flow_key *key,
+ struct sw_flow_mask *mask)
+{
+ memset(match, 0, sizeof(*match));
+ match->key = key;
+ match->mask = mask;
+
+ memset(key, 0, sizeof(*key));
+
+ if (mask) {
+ memset(&mask->key, 0, sizeof(mask->key));
+ mask->range.start = mask->range.end = 0;
+ }
+}
+
+static bool ovs_match_validate(const struct sw_flow_match *match,
+ u64 key_attrs, u64 mask_attrs)
+{
+ u64 key_expected = 1 << OVS_KEY_ATTR_ETHERNET;
+ u64 mask_allowed = key_attrs; /* At most allow all key attributes */
+
+ /* The following mask attributes allowed only if they
+ * pass the validation tests. */
+ mask_allowed &= ~((1 << OVS_KEY_ATTR_IPV4)
+ | (1 << OVS_KEY_ATTR_IPV6)
+ | (1 << OVS_KEY_ATTR_TCP)
+ | (1 << OVS_KEY_ATTR_UDP)
+ | (1 << OVS_KEY_ATTR_ICMP)
+ | (1 << OVS_KEY_ATTR_ICMPV6)
+ | (1 << OVS_KEY_ATTR_ARP)
+ | (1 << OVS_KEY_ATTR_ND));
+
+ /* Always allowed mask fields. */
+ mask_allowed |= ((1 << OVS_KEY_ATTR_TUNNEL)
+ | (1 << OVS_KEY_ATTR_IN_PORT)
+ | (1 << OVS_KEY_ATTR_ETHERTYPE));
+
+ /* Check key attributes. */
+ if (match->key->eth.type == htons(ETH_P_ARP)
+ || match->key->eth.type == htons(ETH_P_RARP)) {
+ key_expected |= 1 << OVS_KEY_ATTR_ARP;
+ if (match->mask && (match->mask->key.eth.type == htons(0xffff)))
+ mask_allowed |= 1 << OVS_KEY_ATTR_ARP;
+ }
+
+ if (match->key->eth.type == htons(ETH_P_IP)) {
+ key_expected |= 1 << OVS_KEY_ATTR_IPV4;
+ if (match->mask && (match->mask->key.eth.type == htons(0xffff)))
+ mask_allowed |= 1 << OVS_KEY_ATTR_IPV4;
+
+ if (match->key->ip.frag != OVS_FRAG_TYPE_LATER) {
+ if (match->key->ip.proto == IPPROTO_UDP) {
+ key_expected |= 1 << OVS_KEY_ATTR_UDP;
+ if (match->mask && (match->mask->key.ip.proto == 0xff))
+ mask_allowed |= 1 << OVS_KEY_ATTR_UDP;
+ }
+
+ if (match->key->ip.proto == IPPROTO_TCP) {
+ key_expected |= 1 << OVS_KEY_ATTR_TCP;
+ if (match->mask && (match->mask->key.ip.proto == 0xff))
+ mask_allowed |= 1 << OVS_KEY_ATTR_TCP;
+ }
+
+ if (match->key->ip.proto == IPPROTO_ICMP) {
+ key_expected |= 1 << OVS_KEY_ATTR_ICMP;
+ if (match->mask && (match->mask->key.ip.proto == 0xff))
+ mask_allowed |= 1 << OVS_KEY_ATTR_ICMP;
+ }
+ }
+ }
+
+ if (match->key->eth.type == htons(ETH_P_IPV6)) {
+ key_expected |= 1 << OVS_KEY_ATTR_IPV6;
+ if (match->mask && (match->mask->key.eth.type == htons(0xffff)))
+ mask_allowed |= 1 << OVS_KEY_ATTR_IPV6;
+
+ if (match->key->ip.frag != OVS_FRAG_TYPE_LATER) {
+ if (match->key->ip.proto == IPPROTO_UDP) {
+ key_expected |= 1 << OVS_KEY_ATTR_UDP;
+ if (match->mask && (match->mask->key.ip.proto == 0xff))
+ mask_allowed |= 1 << OVS_KEY_ATTR_UDP;
+ }
+
+ if (match->key->ip.proto == IPPROTO_TCP) {
+ key_expected |= 1 << OVS_KEY_ATTR_TCP;
+ if (match->mask && (match->mask->key.ip.proto == 0xff))
+ mask_allowed |= 1 << OVS_KEY_ATTR_TCP;
+ }
+
+ if (match->key->ip.proto == IPPROTO_ICMPV6) {
+ key_expected |= 1 << OVS_KEY_ATTR_ICMPV6;
+ if (match->mask && (match->mask->key.ip.proto == 0xff))
+ mask_allowed |= 1 << OVS_KEY_ATTR_ICMPV6;
+
+ if (match->key->ipv6.tp.src ==
+ htons(NDISC_NEIGHBOUR_SOLICITATION) ||
+ match->key->ipv6.tp.src == htons(NDISC_NEIGHBOUR_ADVERTISEMENT)) {
+ key_expected |= 1 << OVS_KEY_ATTR_ND;
+ if (match->mask && (match->mask->key.ipv6.tp.src == htons(0xffff)))
+ mask_allowed |= 1 << OVS_KEY_ATTR_ND;
+ }
+ }
+ }
+ }
+
+ if ((key_attrs & key_expected) != key_expected) {
+ /* Key attributes check failed. */
+ OVS_NLERR("Missing expected key attributes (key_attrs=%llx, expected=%llx).\n",
+ key_attrs, key_expected);
+ return false;
+ }
+
+ if ((mask_attrs & mask_allowed) != mask_attrs) {
+ /* Mask attributes check failed. */
+ OVS_NLERR("Contain more than allowed mask fields (mask_attrs=%llx, mask_allowed=%llx).\n",
+ mask_attrs, mask_allowed);
+ return false;
+ }
+
+ return true;
+}
+
static int check_header(struct sk_buff *skb, int len)
{
if (unlikely(skb->len < len))
@@ -121,12 +299,7 @@ u64 ovs_flow_used_time(unsigned long flow_jiffies)
return cur_ms - idle_ms;
}
-#define SW_FLOW_KEY_OFFSET(field) \
- (offsetof(struct sw_flow_key, field) + \
- FIELD_SIZEOF(struct sw_flow_key, field))
-
-static int parse_ipv6hdr(struct sk_buff *skb, struct sw_flow_key *key,
- int *key_lenp)
+static int parse_ipv6hdr(struct sk_buff *skb, struct sw_flow_key *key)
{
unsigned int nh_ofs = skb_network_offset(skb);
unsigned int nh_len;
@@ -136,8 +309,6 @@ static int parse_ipv6hdr(struct sk_buff *skb, struct sw_flow_key *key,
__be16 frag_off;
int err;
- *key_lenp = SW_FLOW_KEY_OFFSET(ipv6.label);
-
err = check_header(skb, nh_ofs + sizeof(*nh));
if (unlikely(err))
return err;
@@ -176,6 +347,21 @@ static bool icmp6hdr_ok(struct sk_buff *skb)
sizeof(struct icmp6hdr));
}
+void ovs_flow_key_mask(struct sw_flow_key *dst, const struct sw_flow_key *src,
+ const struct sw_flow_mask *mask)
+{
+ u8 *m = (u8 *)&mask->key + mask->range.start;
+ u8 *s = (u8 *)src + mask->range.start;
+ u8 *d = (u8 *)dst + mask->range.start;
+ int i;
+
+ memset(dst, 0, sizeof(*dst));
+ for (i = 0; i < ovs_sw_flow_mask_size_roundup(mask); i++) {
+ *d = *s & *m;
+ d++, s++, m++;
+ }
+}
+
#define TCP_FLAGS_OFFSET 13
#define TCP_FLAG_MASK 0x3f
@@ -224,6 +410,7 @@ struct sw_flow *ovs_flow_alloc(void)
spin_lock_init(&flow->lock);
flow->sf_acts = NULL;
+ flow->mask = NULL;
return flow;
}
@@ -263,7 +450,7 @@ static void free_buckets(struct flex_array *buckets)
flex_array_free(buckets);
}
-struct flow_table *ovs_flow_tbl_alloc(int new_size)
+static struct flow_table *__flow_tbl_alloc(int new_size)
{
struct flow_table *table = kmalloc(sizeof(*table), GFP_KERNEL);
@@ -281,17 +468,15 @@ struct flow_table *ovs_flow_tbl_alloc(int new_size)
table->node_ver = 0;
table->keep_flows = false;
get_random_bytes(&table->hash_seed, sizeof(u32));
+ table->mask_list = NULL;
return table;
}
-void ovs_flow_tbl_destroy(struct flow_table *table)
+static void __flow_tbl_destroy(struct flow_table *table)
{
int i;
- if (!table)
- return;
-
if (table->keep_flows)
goto skip_flows;
@@ -303,31 +488,55 @@ void ovs_flow_tbl_destroy(struct flow_table *table)
hlist_for_each_entry_safe(flow, n, head, hash_node[ver]) {
hlist_del(&flow->hash_node[ver]);
- ovs_flow_free(flow);
+ ovs_flow_free(flow, false);
}
}
+ BUG_ON(!list_empty(table->mask_list));
+ kfree(table->mask_list);
+
skip_flows:
free_buckets(table->buckets);
kfree(table);
}
+struct flow_table *ovs_flow_tbl_alloc(int new_size)
+{
+ struct flow_table *table = __flow_tbl_alloc(new_size);
+
+ if (!table)
+ return NULL;
+
+ table->mask_list = kmalloc(sizeof(struct list_head), GFP_KERNEL);
+ if (!table->mask_list) {
+ table->keep_flows = true;
+ __flow_tbl_destroy(table);
+ return NULL;
+ }
+ INIT_LIST_HEAD(table->mask_list);
+
+ return table;
+}
+
static void flow_tbl_destroy_rcu_cb(struct rcu_head *rcu)
{
struct flow_table *table = container_of(rcu, struct flow_table, rcu);
- ovs_flow_tbl_destroy(table);
+ __flow_tbl_destroy(table);
}
-void ovs_flow_tbl_deferred_destroy(struct flow_table *table)
+void ovs_flow_tbl_destroy(struct flow_table *table, bool deferred)
{
if (!table)
return;
- call_rcu(&table->rcu, flow_tbl_destroy_rcu_cb);
+ if (deferred)
+ call_rcu(&table->rcu, flow_tbl_destroy_rcu_cb);
+ else
+ __flow_tbl_destroy(table);
}
-struct sw_flow *ovs_flow_tbl_next(struct flow_table *table, u32 *bucket, u32 *last)
+struct sw_flow *ovs_flow_dump_next(struct flow_table *table, u32 *bucket, u32 *last)
{
struct sw_flow *flow;
struct hlist_head *head;
@@ -353,11 +562,13 @@ struct sw_flow *ovs_flow_tbl_next(struct flow_table *table, u32 *bucket, u32 *la
return NULL;
}
-static void __flow_tbl_insert(struct flow_table *table, struct sw_flow *flow)
+static void __tbl_insert(struct flow_table *table, struct sw_flow *flow)
{
struct hlist_head *head;
+
head = find_bucket(table, flow->hash);
hlist_add_head_rcu(&flow->hash_node[table->node_ver], head);
+
table->count++;
}
@@ -377,8 +588,10 @@ static void flow_table_copy_flows(struct flow_table *old, struct flow_table *new
head = flex_array_get(old->buckets, i);
hlist_for_each_entry(flow, head, hash_node[old_ver])
- __flow_tbl_insert(new, flow);
+ __tbl_insert(new, flow);
}
+
+ new->mask_list = old->mask_list;
old->keep_flows = true;
}
@@ -386,7 +599,7 @@ static struct flow_table *__flow_tbl_rehash(struct flow_table *table, int n_buck
{
struct flow_table *new_table;
- new_table = ovs_flow_tbl_alloc(n_buckets);
+ new_table = __flow_tbl_alloc(n_buckets);
if (!new_table)
return ERR_PTR(-ENOMEM);
@@ -405,28 +618,30 @@ struct flow_table *ovs_flow_tbl_expand(struct flow_table *table)
return __flow_tbl_rehash(table, table->n_buckets * 2);
}
-void ovs_flow_free(struct sw_flow *flow)
+static void __flow_free(struct sw_flow *flow)
{
- if (unlikely(!flow))
- return;
-
kfree((struct sf_flow_acts __force *)flow->sf_acts);
kmem_cache_free(flow_cache, flow);
}
-/* RCU callback used by ovs_flow_deferred_free. */
static void rcu_free_flow_callback(struct rcu_head *rcu)
{
struct sw_flow *flow = container_of(rcu, struct sw_flow, rcu);
- ovs_flow_free(flow);
+ __flow_free(flow);
}
-/* Schedules 'flow' to be freed after the next RCU grace period.
- * The caller must hold rcu_read_lock for this to be sensible. */
-void ovs_flow_deferred_free(struct sw_flow *flow)
+void ovs_flow_free(struct sw_flow *flow, bool deferred)
{
- call_rcu(&flow->rcu, rcu_free_flow_callback);
+ if (!flow)
+ return;
+
+ ovs_sw_flow_mask_del_ref(flow->mask, deferred);
+
+ if (deferred)
+ call_rcu(&flow->rcu, rcu_free_flow_callback);
+ else
+ __flow_free(flow);
}
/* Schedules 'sf_acts' to be freed after the next RCU grace period.
@@ -497,18 +712,15 @@ static __be16 parse_ethertype(struct sk_buff *skb)
}
static int parse_icmpv6(struct sk_buff *skb, struct sw_flow_key *key,
- int *key_lenp, int nh_len)
+ int nh_len)
{
struct icmp6hdr *icmp = icmp6_hdr(skb);
- int error = 0;
- int key_len;
/* The ICMPv6 type and code fields use the 16-bit transport port
* fields, so we need to store them in 16-bit network byte order.
*/
key->ipv6.tp.src = htons(icmp->icmp6_type);
key->ipv6.tp.dst = htons(icmp->icmp6_code);
- key_len = SW_FLOW_KEY_OFFSET(ipv6.tp);
if (icmp->icmp6_code == 0 &&
(icmp->icmp6_type == NDISC_NEIGHBOUR_SOLICITATION ||
@@ -517,21 +729,17 @@ static int parse_icmpv6(struct sk_buff *skb, struct sw_flow_key *key,
struct nd_msg *nd;
int offset;
- key_len = SW_FLOW_KEY_OFFSET(ipv6.nd);
-
/* In order to process neighbor discovery options, we need the
* entire packet.
*/
if (unlikely(icmp_len < sizeof(*nd)))
- goto out;
- if (unlikely(skb_linearize(skb))) {
- error = -ENOMEM;
- goto out;
- }
+ return 0;
+
+ if (unlikely(skb_linearize(skb)))
+ return -ENOMEM;
nd = (struct nd_msg *)skb_transport_header(skb);
key->ipv6.nd.target = nd->target;
- key_len = SW_FLOW_KEY_OFFSET(ipv6.nd);
icmp_len -= sizeof(*nd);
offset = 0;
@@ -541,7 +749,7 @@ static int parse_icmpv6(struct sk_buff *skb, struct sw_flow_key *key,
int opt_len = nd_opt->nd_opt_len * 8;
if (unlikely(!opt_len || opt_len > icmp_len))
- goto invalid;
+ return 0;
/* Store the link layer address if the appropriate
* option is provided. It is considered an error if
@@ -566,16 +774,14 @@ static int parse_icmpv6(struct sk_buff *skb, struct sw_flow_key *key,
}
}
- goto out;
+ return 0;
invalid:
memset(&key->ipv6.nd.target, 0, sizeof(key->ipv6.nd.target));
memset(key->ipv6.nd.sll, 0, sizeof(key->ipv6.nd.sll));
memset(key->ipv6.nd.tll, 0, sizeof(key->ipv6.nd.tll));
-out:
- *key_lenp = key_len;
- return error;
+ return 0;
}
/**
@@ -584,7 +790,6 @@ out:
* Ethernet header
* @in_port: port number on which @skb was received.
* @key: output flow key
- * @key_lenp: length of output flow key
*
* The caller must ensure that skb->len >= ETH_HLEN.
*
@@ -602,11 +807,9 @@ out:
* of a correct length, otherwise the same as skb->network_header.
* For other key->eth.type values it is left untouched.
*/
-int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key,
- int *key_lenp)
+int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key)
{
- int error = 0;
- int key_len = SW_FLOW_KEY_OFFSET(eth);
+ int error;
struct ethhdr *eth;
memset(key, 0, sizeof(*key));
@@ -649,15 +852,13 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key,
struct iphdr *nh;
__be16 offset;
- key_len = SW_FLOW_KEY_OFFSET(ipv4.addr);
-
error = check_iphdr(skb);
if (unlikely(error)) {
if (error == -EINVAL) {
skb->transport_header = skb->network_header;
error = 0;
}
- goto out;
+ return error;
}
nh = ip_hdr(skb);
@@ -671,7 +872,7 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key,
offset = nh->frag_off & htons(IP_OFFSET);
if (offset) {
key->ip.frag = OVS_FRAG_TYPE_LATER;
- goto out;
+ return 0;
}
if (nh->frag_off & htons(IP_MF) ||
skb_shinfo(skb)->gso_type & SKB_GSO_UDP)
@@ -679,21 +880,18 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key,
/* Transport layer. */
if (key->ip.proto == IPPROTO_TCP) {
- key_len = SW_FLOW_KEY_OFFSET(ipv4.tp);
if (tcphdr_ok(skb)) {
struct tcphdr *tcp = tcp_hdr(skb);
key->ipv4.tp.src = tcp->source;
key->ipv4.tp.dst = tcp->dest;
}
} else if (key->ip.proto == IPPROTO_UDP) {
- key_len = SW_FLOW_KEY_OFFSET(ipv4.tp);
if (udphdr_ok(skb)) {
struct udphdr *udp = udp_hdr(skb);
key->ipv4.tp.src = udp->source;
key->ipv4.tp.dst = udp->dest;
}
} else if (key->ip.proto == IPPROTO_ICMP) {
- key_len = SW_FLOW_KEY_OFFSET(ipv4.tp);
if (icmphdr_ok(skb)) {
struct icmphdr *icmp = icmp_hdr(skb);
/* The ICMP type and code fields use the 16-bit
@@ -722,53 +920,49 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key,
memcpy(&key->ipv4.addr.dst, arp->ar_tip, sizeof(key->ipv4.addr.dst));
memcpy(key->ipv4.arp.sha, arp->ar_sha, ETH_ALEN);
memcpy(key->ipv4.arp.tha, arp->ar_tha, ETH_ALEN);
- key_len = SW_FLOW_KEY_OFFSET(ipv4.arp);
}
} else if (key->eth.type == htons(ETH_P_IPV6)) {
int nh_len; /* IPv6 Header + Extensions */
- nh_len = parse_ipv6hdr(skb, key, &key_len);
+ nh_len = parse_ipv6hdr(skb, key);
if (unlikely(nh_len < 0)) {
- if (nh_len == -EINVAL)
+ if (nh_len == -EINVAL) {
skb->transport_header = skb->network_header;
- else
+ error = 0;
+ } else {
error = nh_len;
- goto out;
+ }
+ return error;
}
if (key->ip.frag == OVS_FRAG_TYPE_LATER)
- goto out;
+ return 0;
if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP)
key->ip.frag = OVS_FRAG_TYPE_FIRST;
/* Transport layer. */
if (key->ip.proto == NEXTHDR_TCP) {
- key_len = SW_FLOW_KEY_OFFSET(ipv6.tp);
if (tcphdr_ok(skb)) {
struct tcphdr *tcp = tcp_hdr(skb);
key->ipv6.tp.src = tcp->source;
key->ipv6.tp.dst = tcp->dest;
}
} else if (key->ip.proto == NEXTHDR_UDP) {
- key_len = SW_FLOW_KEY_OFFSET(ipv6.tp);
if (udphdr_ok(skb)) {
struct udphdr *udp = udp_hdr(skb);
key->ipv6.tp.src = udp->source;
key->ipv6.tp.dst = udp->dest;
}
} else if (key->ip.proto == NEXTHDR_ICMP) {
- key_len = SW_FLOW_KEY_OFFSET(ipv6.tp);
if (icmp6hdr_ok(skb)) {
- error = parse_icmpv6(skb, key, &key_len, nh_len);
- if (error < 0)
- goto out;
+ error = parse_icmpv6(skb, key, nh_len);
+ if (error)
+ return error;
}
}
}
-out:
- *key_lenp = key_len;
- return error;
+ return 0;
}
static u32 ovs_flow_hash(const struct sw_flow_key *key, int key_start, int key_len)
@@ -777,7 +971,7 @@ static u32 ovs_flow_hash(const struct sw_flow_key *key, int key_start, int key_l
DIV_ROUND_UP(key_len - key_start, sizeof(u32)), 0);
}
-static int flow_key_start(struct sw_flow_key *key)
+static int flow_key_start(const struct sw_flow_key *key)
{
if (key->tun_key.ipv4_dst)
return 0;
@@ -785,39 +979,95 @@ static int flow_key_start(struct sw_flow_key *key)
return offsetof(struct sw_flow_key, phy);
}
-struct sw_flow *ovs_flow_tbl_lookup(struct flow_table *table,
- struct sw_flow_key *key, int key_len)
+static bool __cmp_key(const struct sw_flow_key *key1,
+ const struct sw_flow_key *key2, int key_start, int key_len)
+{
+ return !memcmp((u8 *)key1 + key_start,
+ (u8 *)key2 + key_start, (key_len - key_start));
+}
+
+static bool __flow_cmp_key(const struct sw_flow *flow,
+ const struct sw_flow_key *key, int key_start, int key_len)
+{
+ return __cmp_key(&flow->key, key, key_start, key_len);
+}
+
+static bool __flow_cmp_unmasked_key(const struct sw_flow *flow,
+ const struct sw_flow_key *key, int key_start, int key_len)
+{
+ return __cmp_key(&flow->unmasked_key, key, key_start, key_len);
+}
+
+bool ovs_flow_cmp_unmasked_key(const struct sw_flow *flow,
+ const struct sw_flow_key *key, int key_len)
+{
+ int key_start;
+ key_start = flow_key_start(key);
+
+ return __flow_cmp_unmasked_key(flow, key, key_start, key_len);
+
+}
+
+struct sw_flow *ovs_flow_lookup_unmasked_key(struct flow_table *table,
+ struct sw_flow_match *match)
+{
+ struct sw_flow_key *unmasked = match->key;
+ int key_len = match->range.end;
+ struct sw_flow *flow;
+
+ flow = ovs_flow_lookup(table, unmasked);
+ if (flow && (!ovs_flow_cmp_unmasked_key(flow, unmasked, key_len)))
+ flow = NULL;
+
+ return flow;
+}
+
+static struct sw_flow *ovs_masked_flow_lookup(struct flow_table *table,
+ const struct sw_flow_key *flow_key,
+ struct sw_flow_mask *mask)
{
struct sw_flow *flow;
struct hlist_head *head;
- u8 *_key;
- int key_start;
+ int key_start = mask->range.start;
+ int key_len = mask->range.end;
u32 hash;
+ struct sw_flow_key masked_key;
- key_start = flow_key_start(key);
- hash = ovs_flow_hash(key, key_start, key_len);
-
- _key = (u8 *) key + key_start;
+ ovs_flow_key_mask(&masked_key, flow_key, mask);
+ hash = ovs_flow_hash(&masked_key, key_start, key_len);
head = find_bucket(table, hash);
hlist_for_each_entry_rcu(flow, head, hash_node[table->node_ver]) {
-
- if (flow->hash == hash &&
- !memcmp((u8 *)&flow->key + key_start, _key, key_len - key_start)) {
+ if (flow->mask == mask &&
+ __flow_cmp_key(flow, &masked_key, key_start, key_len))
return flow;
- }
}
return NULL;
}
-void ovs_flow_tbl_insert(struct flow_table *table, struct sw_flow *flow,
- struct sw_flow_key *key, int key_len)
+struct sw_flow *ovs_flow_lookup(struct flow_table *tbl,
+ const struct sw_flow_key *key)
{
- flow->hash = ovs_flow_hash(key, flow_key_start(key), key_len);
- memcpy(&flow->key, key, sizeof(flow->key));
- __flow_tbl_insert(table, flow);
+ struct sw_flow *flow = NULL;
+ struct sw_flow_mask *mask;
+
+ list_for_each_entry_rcu(mask, tbl->mask_list, list) {
+ flow = ovs_masked_flow_lookup(tbl, key, mask);
+ if (flow) /* Found */
+ break;
+ }
+
+ return flow;
}
-void ovs_flow_tbl_remove(struct flow_table *table, struct sw_flow *flow)
+
+void ovs_flow_insert(struct flow_table *table, struct sw_flow *flow)
+{
+ flow->hash = ovs_flow_hash(&flow->key, flow->mask->range.start,
+ flow->mask->range.end);
+ __tbl_insert(table, flow);
+}
+
+void ovs_flow_remove(struct flow_table *table, struct sw_flow *flow)
{
BUG_ON(table->count == 0);
hlist_del_rcu(&flow->hash_node[table->node_ver]);
@@ -844,149 +1094,84 @@ const int ovs_key_lens[OVS_KEY_ATTR_MAX + 1] = {
[OVS_KEY_ATTR_TUNNEL] = -1,
};
-static int ipv4_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_len,
- const struct nlattr *a[], u32 *attrs)
-{
- const struct ovs_key_icmp *icmp_key;
- const struct ovs_key_tcp *tcp_key;
- const struct ovs_key_udp *udp_key;
-
- switch (swkey->ip.proto) {
- case IPPROTO_TCP:
- if (!(*attrs & (1 << OVS_KEY_ATTR_TCP)))
- return -EINVAL;
- *attrs &= ~(1 << OVS_KEY_ATTR_TCP);
-
- *key_len = SW_FLOW_KEY_OFFSET(ipv4.tp);
- tcp_key = nla_data(a[OVS_KEY_ATTR_TCP]);
- swkey->ipv4.tp.src = tcp_key->tcp_src;
- swkey->ipv4.tp.dst = tcp_key->tcp_dst;
- break;
-
- case IPPROTO_UDP:
- if (!(*attrs & (1 << OVS_KEY_ATTR_UDP)))
- return -EINVAL;
- *attrs &= ~(1 << OVS_KEY_ATTR_UDP);
-
- *key_len = SW_FLOW_KEY_OFFSET(ipv4.tp);
- udp_key = nla_data(a[OVS_KEY_ATTR_UDP]);
- swkey->ipv4.tp.src = udp_key->udp_src;
- swkey->ipv4.tp.dst = udp_key->udp_dst;
- break;
-
- case IPPROTO_ICMP:
- if (!(*attrs & (1 << OVS_KEY_ATTR_ICMP)))
- return -EINVAL;
- *attrs &= ~(1 << OVS_KEY_ATTR_ICMP);
-
- *key_len = SW_FLOW_KEY_OFFSET(ipv4.tp);
- icmp_key = nla_data(a[OVS_KEY_ATTR_ICMP]);
- swkey->ipv4.tp.src = htons(icmp_key->icmp_type);
- swkey->ipv4.tp.dst = htons(icmp_key->icmp_code);
- break;
- }
-
- return 0;
-}
-
-static int ipv6_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_len,
- const struct nlattr *a[], u32 *attrs)
+static bool is_all_zero(const u8 *fp, size_t size)
{
- const struct ovs_key_icmpv6 *icmpv6_key;
- const struct ovs_key_tcp *tcp_key;
- const struct ovs_key_udp *udp_key;
-
- switch (swkey->ip.proto) {
- case IPPROTO_TCP:
- if (!(*attrs & (1 << OVS_KEY_ATTR_TCP)))
- return -EINVAL;
- *attrs &= ~(1 << OVS_KEY_ATTR_TCP);
-
- *key_len = SW_FLOW_KEY_OFFSET(ipv6.tp);
- tcp_key = nla_data(a[OVS_KEY_ATTR_TCP]);
- swkey->ipv6.tp.src = tcp_key->tcp_src;
- swkey->ipv6.tp.dst = tcp_key->tcp_dst;
- break;
-
- case IPPROTO_UDP:
- if (!(*attrs & (1 << OVS_KEY_ATTR_UDP)))
- return -EINVAL;
- *attrs &= ~(1 << OVS_KEY_ATTR_UDP);
-
- *key_len = SW_FLOW_KEY_OFFSET(ipv6.tp);
- udp_key = nla_data(a[OVS_KEY_ATTR_UDP]);
- swkey->ipv6.tp.src = udp_key->udp_src;
- swkey->ipv6.tp.dst = udp_key->udp_dst;
- break;
-
- case IPPROTO_ICMPV6:
- if (!(*attrs & (1 << OVS_KEY_ATTR_ICMPV6)))
- return -EINVAL;
- *attrs &= ~(1 << OVS_KEY_ATTR_ICMPV6);
-
- *key_len = SW_FLOW_KEY_OFFSET(ipv6.tp);
- icmpv6_key = nla_data(a[OVS_KEY_ATTR_ICMPV6]);
- swkey->ipv6.tp.src = htons(icmpv6_key->icmpv6_type);
- swkey->ipv6.tp.dst = htons(icmpv6_key->icmpv6_code);
+ int i;
- if (swkey->ipv6.tp.src == htons(NDISC_NEIGHBOUR_SOLICITATION) ||
- swkey->ipv6.tp.src == htons(NDISC_NEIGHBOUR_ADVERTISEMENT)) {
- const struct ovs_key_nd *nd_key;
+ if (!fp)
+ return false;
- if (!(*attrs & (1 << OVS_KEY_ATTR_ND)))
- return -EINVAL;
- *attrs &= ~(1 << OVS_KEY_ATTR_ND);
-
- *key_len = SW_FLOW_KEY_OFFSET(ipv6.nd);
- nd_key = nla_data(a[OVS_KEY_ATTR_ND]);
- memcpy(&swkey->ipv6.nd.target, nd_key->nd_target,
- sizeof(swkey->ipv6.nd.target));
- memcpy(swkey->ipv6.nd.sll, nd_key->nd_sll, ETH_ALEN);
- memcpy(swkey->ipv6.nd.tll, nd_key->nd_tll, ETH_ALEN);
- }
- break;
- }
+ for (i = 0; i < size; i++)
+ if (fp[i])
+ return false;
- return 0;
+ return true;
}
-static int parse_flow_nlattrs(const struct nlattr *attr,
- const struct nlattr *a[], u32 *attrsp)
+static int __parse_flow_nlattrs(const struct nlattr *attr,
+ const struct nlattr *a[],
+ u64 *attrsp, bool nz)
{
const struct nlattr *nla;
u32 attrs;
int rem;
- attrs = 0;
+ attrs = *attrsp;
nla_for_each_nested(nla, attr, rem) {
u16 type = nla_type(nla);
int expected_len;
- if (type > OVS_KEY_ATTR_MAX || attrs & (1 << type))
+ if (type > OVS_KEY_ATTR_MAX) {
+ OVS_NLERR("Unknown key attribute (type=%d, max=%d).\n",
+ type, OVS_KEY_ATTR_MAX);
+ }
+
+ if (attrs & (1 << type)) {
+ OVS_NLERR("Duplicate key attribute (type %d).\n", type);
return -EINVAL;
+ }
expected_len = ovs_key_lens[type];
- if (nla_len(nla) != expected_len && expected_len != -1)
+ if (nla_len(nla) != expected_len && expected_len != -1) {
+ OVS_NLERR("Key attribute has unexpected length (type=%d"
+ ", length=%d, expected=%d).\n", type,
+ nla_len(nla), expected_len);
return -EINVAL;
+ }
- attrs |= 1 << type;
- a[type] = nla;
+ if (!nz || !is_all_zero(nla_data(nla), expected_len)) {
+ attrs |= 1 << type;
+ a[type] = nla;
+ }
}
- if (rem)
+ if (rem) {
+ OVS_NLERR("Message has %d unknown bytes.\n", rem);
return -EINVAL;
+ }
*attrsp = attrs;
return 0;
}
+static int parse_flow_mask_nlattrs(const struct nlattr *attr,
+ const struct nlattr *a[], u64 *attrsp)
+{
+ return __parse_flow_nlattrs(attr, a, attrsp, true);
+}
+
+static int parse_flow_nlattrs(const struct nlattr *attr,
+ const struct nlattr *a[], u64 *attrsp)
+{
+ return __parse_flow_nlattrs(attr, a, attrsp, false);
+}
+
int ovs_ipv4_tun_from_nlattr(const struct nlattr *attr,
- struct ovs_key_ipv4_tunnel *tun_key)
+ struct sw_flow_match *match, bool is_mask)
{
struct nlattr *a;
int rem;
bool ttl = false;
-
- memset(tun_key, 0, sizeof(*tun_key));
+ __be16 tun_flags = 0;
nla_for_each_nested(a, attr, rem) {
int type = nla_type(a);
@@ -1000,53 +1185,78 @@ int ovs_ipv4_tun_from_nlattr(const struct nlattr *attr,
[OVS_TUNNEL_KEY_ATTR_CSUM] = 0,
};
- if (type > OVS_TUNNEL_KEY_ATTR_MAX ||
- ovs_tunnel_key_lens[type] != nla_len(a))
+ if (type > OVS_TUNNEL_KEY_ATTR_MAX) {
+ OVS_NLERR("Unknown IPv4 tunnel attribute (type=%d, max=%d).\n",
+ type, OVS_TUNNEL_KEY_ATTR_MAX);
return -EINVAL;
+ }
+
+ if (ovs_tunnel_key_lens[type] != nla_len(a)) {
+ OVS_NLERR("IPv4 tunnel attribute type has unexpected "
+ " length (type=%d, length=%d, expected=%d).\n",
+ type, nla_len(a), ovs_tunnel_key_lens[type]);
+ return -EINVAL;
+ }
switch (type) {
case OVS_TUNNEL_KEY_ATTR_ID:
- tun_key->tun_id = nla_get_be64(a);
- tun_key->tun_flags |= TUNNEL_KEY;
+ SW_FLOW_KEY_PUT(match, tun_key.tun_id,
+ nla_get_be64(a), is_mask);
+ tun_flags |= TUNNEL_KEY;
break;
case OVS_TUNNEL_KEY_ATTR_IPV4_SRC:
- tun_key->ipv4_src = nla_get_be32(a);
+ SW_FLOW_KEY_PUT(match, tun_key.ipv4_src,
+ nla_get_be32(a), is_mask);
break;
case OVS_TUNNEL_KEY_ATTR_IPV4_DST:
- tun_key->ipv4_dst = nla_get_be32(a);
+ SW_FLOW_KEY_PUT(match, tun_key.ipv4_dst,
+ nla_get_be32(a), is_mask);
break;
case OVS_TUNNEL_KEY_ATTR_TOS:
- tun_key->ipv4_tos = nla_get_u8(a);
+ SW_FLOW_KEY_PUT(match, tun_key.ipv4_tos,
+ nla_get_u8(a), is_mask);
break;
case OVS_TUNNEL_KEY_ATTR_TTL:
- tun_key->ipv4_ttl = nla_get_u8(a);
+ SW_FLOW_KEY_PUT(match, tun_key.ipv4_ttl,
+ nla_get_u8(a), is_mask);
ttl = true;
break;
case OVS_TUNNEL_KEY_ATTR_DONT_FRAGMENT:
- tun_key->tun_flags |= TUNNEL_DONT_FRAGMENT;
+ tun_flags |= TUNNEL_DONT_FRAGMENT;
break;
case OVS_TUNNEL_KEY_ATTR_CSUM:
- tun_key->tun_flags |= TUNNEL_CSUM;
+ tun_flags |= TUNNEL_CSUM;
break;
default:
return -EINVAL;
-
}
}
- if (rem > 0)
- return -EINVAL;
- if (!tun_key->ipv4_dst)
- return -EINVAL;
+ SW_FLOW_KEY_PUT(match, tun_key.tun_flags, tun_flags, is_mask);
- if (!ttl)
+ if (rem > 0) {
+ OVS_NLERR("IPv4 tunnel attribute has %d unknown bytes.\n", rem);
return -EINVAL;
+ }
+
+ if (!is_mask) {
+ if (!match->key->tun_key.ipv4_dst) {
+ OVS_NLERR("IPv4 tunnel destination address is zero.\n");
+ return -EINVAL;
+ }
+
+ if (!ttl) {
+ OVS_NLERR("IPv4 tunnel TTL not specified.\n");
+ return -EINVAL;
+ }
+ }
return 0;
}
int ovs_ipv4_tun_to_nlattr(struct sk_buff *skb,
- const struct ovs_key_ipv4_tunnel *tun_key)
+ const struct ovs_key_ipv4_tunnel *tun_key,
+ const struct ovs_key_ipv4_tunnel *output)
{
struct nlattr *nla;
@@ -1054,23 +1264,24 @@ int ovs_ipv4_tun_to_nlattr(struct sk_buff *skb,
if (!nla)
return -EMSGSIZE;
- if (tun_key->tun_flags & TUNNEL_KEY &&
- nla_put_be64(skb, OVS_TUNNEL_KEY_ATTR_ID, tun_key->tun_id))
+ if (output->tun_flags & TUNNEL_KEY &&
+ nla_put_be64(skb, OVS_TUNNEL_KEY_ATTR_ID, output->tun_id))
return -EMSGSIZE;
- if (tun_key->ipv4_src &&
- nla_put_be32(skb, OVS_TUNNEL_KEY_ATTR_IPV4_SRC, tun_key->ipv4_src))
+ if (output->ipv4_src &&
+ nla_put_be32(skb, OVS_TUNNEL_KEY_ATTR_IPV4_SRC, output->ipv4_src))
return -EMSGSIZE;
- if (nla_put_be32(skb, OVS_TUNNEL_KEY_ATTR_IPV4_DST, tun_key->ipv4_dst))
+ if (output->ipv4_dst &&
+ nla_put_be32(skb, OVS_TUNNEL_KEY_ATTR_IPV4_DST, output->ipv4_dst))
return -EMSGSIZE;
- if (tun_key->ipv4_tos &&
- nla_put_u8(skb, OVS_TUNNEL_KEY_ATTR_TOS, tun_key->ipv4_tos))
+ if (output->ipv4_tos &&
+ nla_put_u8(skb, OVS_TUNNEL_KEY_ATTR_TOS, output->ipv4_tos))
return -EMSGSIZE;
- if (nla_put_u8(skb, OVS_TUNNEL_KEY_ATTR_TTL, tun_key->ipv4_ttl))
+ if (nla_put_u8(skb, OVS_TUNNEL_KEY_ATTR_TTL, output->ipv4_ttl))
return -EMSGSIZE;
- if ((tun_key->tun_flags & TUNNEL_DONT_FRAGMENT) &&
+ if ((output->tun_flags & TUNNEL_DONT_FRAGMENT) &&
nla_put_flag(skb, OVS_TUNNEL_KEY_ATTR_DONT_FRAGMENT))
return -EMSGSIZE;
- if ((tun_key->tun_flags & TUNNEL_CSUM) &&
+ if ((output->tun_flags & TUNNEL_CSUM) &&
nla_put_flag(skb, OVS_TUNNEL_KEY_ATTR_CSUM))
return -EMSGSIZE;
@@ -1078,176 +1289,372 @@ int ovs_ipv4_tun_to_nlattr(struct sk_buff *skb,
return 0;
}
-/**
- * ovs_flow_from_nlattrs - parses Netlink attributes into a flow key.
- * @swkey: receives the extracted flow key.
- * @key_lenp: number of bytes used in @swkey.
- * @attr: Netlink attribute holding nested %OVS_KEY_ATTR_* Netlink attribute
- * sequence.
- */
-int ovs_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp,
- const struct nlattr *attr)
+static int metadata_from_nlattrs(struct sw_flow_match *match, u64 *attrs,
+ const struct nlattr **a, bool is_mask)
{
- const struct nlattr *a[OVS_KEY_ATTR_MAX + 1];
- const struct ovs_key_ethernet *eth_key;
- int key_len;
- u32 attrs;
- int err;
+ if (*attrs & (1 << OVS_KEY_ATTR_PRIORITY)) {
+ SW_FLOW_KEY_PUT(match, phy.priority,
+ nla_get_u32(a[OVS_KEY_ATTR_PRIORITY]), is_mask);
+ *attrs &= ~(1 << OVS_KEY_ATTR_PRIORITY);
+ }
- memset(swkey, 0, sizeof(struct sw_flow_key));
- key_len = SW_FLOW_KEY_OFFSET(eth);
+ if (*attrs & (1 << OVS_KEY_ATTR_IN_PORT)) {
+ u32 in_port = nla_get_u32(a[OVS_KEY_ATTR_IN_PORT]);
- err = parse_flow_nlattrs(attr, a, &attrs);
- if (err)
- return err;
+ if (is_mask)
+ in_port = 0xffffffff; /* Always exact match in_port. */
+ else if (in_port >= DP_MAX_PORTS)
+ return -EINVAL;
- /* Metadata attributes. */
- if (attrs & (1 << OVS_KEY_ATTR_PRIORITY)) {
- swkey->phy.priority = nla_get_u32(a[OVS_KEY_ATTR_PRIORITY]);
- attrs &= ~(1 << OVS_KEY_ATTR_PRIORITY);
+ SW_FLOW_KEY_PUT(match, phy.in_port, in_port, is_mask);
+ *attrs &= ~(1 << OVS_KEY_ATTR_IN_PORT);
+ } else if (!is_mask) {
+ SW_FLOW_KEY_PUT(match, phy.in_port, DP_MAX_PORTS, is_mask);
}
- if (attrs & (1 << OVS_KEY_ATTR_IN_PORT)) {
- u32 in_port = nla_get_u32(a[OVS_KEY_ATTR_IN_PORT]);
- if (in_port >= DP_MAX_PORTS)
- return -EINVAL;
- swkey->phy.in_port = in_port;
- attrs &= ~(1 << OVS_KEY_ATTR_IN_PORT);
- } else {
- swkey->phy.in_port = DP_MAX_PORTS;
+
+ if (*attrs & (1 << OVS_KEY_ATTR_SKB_MARK)) {
+ uint32_t mark = nla_get_u32(a[OVS_KEY_ATTR_SKB_MARK]);
+
+ SW_FLOW_KEY_PUT(match, phy.skb_mark, mark, is_mask);
+ *attrs &= ~(1 << OVS_KEY_ATTR_SKB_MARK);
}
- if (attrs & (1 << OVS_KEY_ATTR_SKB_MARK)) {
- swkey->phy.skb_mark = nla_get_u32(a[OVS_KEY_ATTR_SKB_MARK]);
- attrs &= ~(1 << OVS_KEY_ATTR_SKB_MARK);
+ if (*attrs & (1 << OVS_KEY_ATTR_TUNNEL)) {
+ if (ovs_ipv4_tun_from_nlattr(a[OVS_KEY_ATTR_TUNNEL], match,
+ is_mask))
+ return -EINVAL;
+ *attrs &= ~(1 << OVS_KEY_ATTR_TUNNEL);
}
+ return 0;
+}
- if (attrs & (1 << OVS_KEY_ATTR_TUNNEL)) {
- err = ovs_ipv4_tun_from_nlattr(a[OVS_KEY_ATTR_TUNNEL], &swkey->tun_key);
- if (err)
- return err;
+static int ovs_key_from_nlattrs(struct sw_flow_match *match, u64 attrs,
+ const struct nlattr **a, bool is_mask)
+{
+ int err;
+ u64 orig_attrs = attrs;
- attrs &= ~(1 << OVS_KEY_ATTR_TUNNEL);
- }
+ err = metadata_from_nlattrs(match, &attrs, a, is_mask);
+ if (err)
+ return err;
- /* Data attributes. */
- if (!(attrs & (1 << OVS_KEY_ATTR_ETHERNET)))
- return -EINVAL;
- attrs &= ~(1 << OVS_KEY_ATTR_ETHERNET);
+ if (attrs & (1 << OVS_KEY_ATTR_ETHERNET)) {
+ const struct ovs_key_ethernet *eth_key;
- eth_key = nla_data(a[OVS_KEY_ATTR_ETHERNET]);
- memcpy(swkey->eth.src, eth_key->eth_src, ETH_ALEN);
- memcpy(swkey->eth.dst, eth_key->eth_dst, ETH_ALEN);
+ eth_key = nla_data(a[OVS_KEY_ATTR_ETHERNET]);
+ SW_FLOW_KEY_MEMCPY(match, eth.src,
+ eth_key->eth_src, ETH_ALEN, is_mask);
+ SW_FLOW_KEY_MEMCPY(match, eth.dst,
+ eth_key->eth_dst, ETH_ALEN, is_mask);
+ attrs &= ~(1 << OVS_KEY_ATTR_ETHERNET);
+ }
- if (attrs & (1u << OVS_KEY_ATTR_ETHERTYPE) &&
- nla_get_be16(a[OVS_KEY_ATTR_ETHERTYPE]) == htons(ETH_P_8021Q)) {
- const struct nlattr *encap;
+ if (attrs & (1 << OVS_KEY_ATTR_VLAN)) {
__be16 tci;
- if (attrs != ((1 << OVS_KEY_ATTR_VLAN) |
- (1 << OVS_KEY_ATTR_ETHERTYPE) |
- (1 << OVS_KEY_ATTR_ENCAP)))
- return -EINVAL;
-
- encap = a[OVS_KEY_ATTR_ENCAP];
tci = nla_get_be16(a[OVS_KEY_ATTR_VLAN]);
- if (tci & htons(VLAN_TAG_PRESENT)) {
- swkey->eth.tci = tci;
-
- err = parse_flow_nlattrs(encap, a, &attrs);
- if (err)
- return err;
- } else if (!tci) {
- /* Corner case for truncated 802.1Q header. */
- if (nla_len(encap))
- return -EINVAL;
+ if (!(tci & htons(VLAN_TAG_PRESENT))) {
+ if (is_mask)
+ OVS_NLERR("VLAN TCI mask does not have exact match for VLAN_TAG_PRESENT bit.\n");
+ else
+ OVS_NLERR("VLAN TCI does not have VLAN_TAG_PRESENT bit set.\n");
- swkey->eth.type = htons(ETH_P_8021Q);
- *key_lenp = key_len;
- return 0;
- } else {
return -EINVAL;
}
- }
+
+ SW_FLOW_KEY_PUT(match, eth.tci, tci, is_mask);
+ attrs &= ~(1 << OVS_KEY_ATTR_VLAN);
+ } else if (!is_mask)
+ SW_FLOW_KEY_PUT(match, eth.tci, htons(0xffff), true);
if (attrs & (1 << OVS_KEY_ATTR_ETHERTYPE)) {
- swkey->eth.type = nla_get_be16(a[OVS_KEY_ATTR_ETHERTYPE]);
- if (ntohs(swkey->eth.type) < ETH_P_802_3_MIN)
+ __be16 eth_type;
+
+ eth_type = nla_get_be16(a[OVS_KEY_ATTR_ETHERTYPE]);
+ if (is_mask) {
+ /* Always exact match EtherType. */
+ eth_type = htons(0xffff);
+ } else if (ntohs(eth_type) < ETH_P_802_3_MIN) {
+ OVS_NLERR("EtherType is less than minimum (type=%x, min=%x).\n",
+ ntohs(eth_type), ETH_P_802_3_MIN);
return -EINVAL;
+ }
+
+ SW_FLOW_KEY_PUT(match, eth.type, eth_type, is_mask);
attrs &= ~(1 << OVS_KEY_ATTR_ETHERTYPE);
- } else {
- swkey->eth.type = htons(ETH_P_802_2);
+ } else if (!is_mask) {
+ SW_FLOW_KEY_PUT(match, eth.type, htons(ETH_P_802_2), is_mask);
}
- if (swkey->eth.type == htons(ETH_P_IP)) {
+ if (attrs & (1 << OVS_KEY_ATTR_IPV4)) {
const struct ovs_key_ipv4 *ipv4_key;
- if (!(attrs & (1 << OVS_KEY_ATTR_IPV4)))
- return -EINVAL;
- attrs &= ~(1 << OVS_KEY_ATTR_IPV4);
-
- key_len = SW_FLOW_KEY_OFFSET(ipv4.addr);
ipv4_key = nla_data(a[OVS_KEY_ATTR_IPV4]);
- if (ipv4_key->ipv4_frag > OVS_FRAG_TYPE_MAX)
+ if (!is_mask && ipv4_key->ipv4_frag > OVS_FRAG_TYPE_MAX) {
+ OVS_NLERR("Unknown IPv4 fragment type (value=%d, max=%d).\n",
+ ipv4_key->ipv4_frag, OVS_FRAG_TYPE_MAX);
return -EINVAL;
- swkey->ip.proto = ipv4_key->ipv4_proto;
- swkey->ip.tos = ipv4_key->ipv4_tos;
- swkey->ip.ttl = ipv4_key->ipv4_ttl;
- swkey->ip.frag = ipv4_key->ipv4_frag;
- swkey->ipv4.addr.src = ipv4_key->ipv4_src;
- swkey->ipv4.addr.dst = ipv4_key->ipv4_dst;
-
- if (swkey->ip.frag != OVS_FRAG_TYPE_LATER) {
- err = ipv4_flow_from_nlattrs(swkey, &key_len, a, &attrs);
- if (err)
- return err;
}
- } else if (swkey->eth.type == htons(ETH_P_IPV6)) {
- const struct ovs_key_ipv6 *ipv6_key;
+ SW_FLOW_KEY_PUT(match, ip.proto,
+ ipv4_key->ipv4_proto, is_mask);
+ SW_FLOW_KEY_PUT(match, ip.tos,
+ ipv4_key->ipv4_tos, is_mask);
+ SW_FLOW_KEY_PUT(match, ip.ttl,
+ ipv4_key->ipv4_ttl, is_mask);
+ SW_FLOW_KEY_PUT(match, ip.frag,
+ ipv4_key->ipv4_frag, is_mask);
+ SW_FLOW_KEY_PUT(match, ipv4.addr.src,
+ ipv4_key->ipv4_src, is_mask);
+ SW_FLOW_KEY_PUT(match, ipv4.addr.dst,
+ ipv4_key->ipv4_dst, is_mask);
+ attrs &= ~(1 << OVS_KEY_ATTR_IPV4);
+ }
- if (!(attrs & (1 << OVS_KEY_ATTR_IPV6)))
- return -EINVAL;
- attrs &= ~(1 << OVS_KEY_ATTR_IPV6);
+ if (attrs & (1 << OVS_KEY_ATTR_IPV6)) {
+ const struct ovs_key_ipv6 *ipv6_key;
- key_len = SW_FLOW_KEY_OFFSET(ipv6.label);
ipv6_key = nla_data(a[OVS_KEY_ATTR_IPV6]);
- if (ipv6_key->ipv6_frag > OVS_FRAG_TYPE_MAX)
+ if (!is_mask && ipv6_key->ipv6_frag > OVS_FRAG_TYPE_MAX) {
+ OVS_NLERR("Unknown IPv6 fragment type (value=%d, max=%d).\n",
+ ipv6_key->ipv6_frag, OVS_FRAG_TYPE_MAX);
return -EINVAL;
- swkey->ipv6.label = ipv6_key->ipv6_label;
- swkey->ip.proto = ipv6_key->ipv6_proto;
- swkey->ip.tos = ipv6_key->ipv6_tclass;
- swkey->ip.ttl = ipv6_key->ipv6_hlimit;
- swkey->ip.frag = ipv6_key->ipv6_frag;
- memcpy(&swkey->ipv6.addr.src, ipv6_key->ipv6_src,
- sizeof(swkey->ipv6.addr.src));
- memcpy(&swkey->ipv6.addr.dst, ipv6_key->ipv6_dst,
- sizeof(swkey->ipv6.addr.dst));
-
- if (swkey->ip.frag != OVS_FRAG_TYPE_LATER) {
- err = ipv6_flow_from_nlattrs(swkey, &key_len, a, &attrs);
- if (err)
- return err;
}
- } else if (swkey->eth.type == htons(ETH_P_ARP) ||
- swkey->eth.type == htons(ETH_P_RARP)) {
+ SW_FLOW_KEY_PUT(match, ipv6.label,
+ ipv6_key->ipv6_label, is_mask);
+ SW_FLOW_KEY_PUT(match, ip.proto,
+ ipv6_key->ipv6_proto, is_mask);
+ SW_FLOW_KEY_PUT(match, ip.tos,
+ ipv6_key->ipv6_tclass, is_mask);
+ SW_FLOW_KEY_PUT(match, ip.ttl,
+ ipv6_key->ipv6_hlimit, is_mask);
+ SW_FLOW_KEY_PUT(match, ip.frag,
+ ipv6_key->ipv6_frag, is_mask);
+ SW_FLOW_KEY_MEMCPY(match, ipv6.addr.src,
+ ipv6_key->ipv6_src,
+ sizeof(match->key->ipv6.addr.src),
+ is_mask);
+ SW_FLOW_KEY_MEMCPY(match, ipv6.addr.dst,
+ ipv6_key->ipv6_dst,
+ sizeof(match->key->ipv6.addr.dst),
+ is_mask);
+
+ attrs &= ~(1 << OVS_KEY_ATTR_IPV6);
+ }
+
+ if (attrs & (1 << OVS_KEY_ATTR_ARP)) {
const struct ovs_key_arp *arp_key;
- if (!(attrs & (1 << OVS_KEY_ATTR_ARP)))
+ arp_key = nla_data(a[OVS_KEY_ATTR_ARP]);
+ if (!is_mask && (arp_key->arp_op & htons(0xff00))) {
+ OVS_NLERR("Unknown ARP opcode (opcode=%d).\n",
+ arp_key->arp_op);
return -EINVAL;
+ }
+
+ SW_FLOW_KEY_PUT(match, ipv4.addr.src,
+ arp_key->arp_sip, is_mask);
+ SW_FLOW_KEY_PUT(match, ipv4.addr.dst,
+ arp_key->arp_tip, is_mask);
+ SW_FLOW_KEY_PUT(match, ip.proto,
+ ntohs(arp_key->arp_op), is_mask);
+ SW_FLOW_KEY_MEMCPY(match, ipv4.arp.sha,
+ arp_key->arp_sha, ETH_ALEN, is_mask);
+ SW_FLOW_KEY_MEMCPY(match, ipv4.arp.tha,
+ arp_key->arp_tha, ETH_ALEN, is_mask);
+
attrs &= ~(1 << OVS_KEY_ATTR_ARP);
+ }
- key_len = SW_FLOW_KEY_OFFSET(ipv4.arp);
- arp_key = nla_data(a[OVS_KEY_ATTR_ARP]);
- swkey->ipv4.addr.src = arp_key->arp_sip;
- swkey->ipv4.addr.dst = arp_key->arp_tip;
- if (arp_key->arp_op & htons(0xff00))
+ if (attrs & (1 << OVS_KEY_ATTR_TCP)) {
+ const struct ovs_key_tcp *tcp_key;
+
+ tcp_key = nla_data(a[OVS_KEY_ATTR_TCP]);
+ if (orig_attrs & (1 << OVS_KEY_ATTR_IPV4)) {
+ SW_FLOW_KEY_PUT(match, ipv4.tp.src,
+ tcp_key->tcp_src, is_mask);
+ SW_FLOW_KEY_PUT(match, ipv4.tp.dst,
+ tcp_key->tcp_dst, is_mask);
+ } else {
+ SW_FLOW_KEY_PUT(match, ipv6.tp.src,
+ tcp_key->tcp_src, is_mask);
+ SW_FLOW_KEY_PUT(match, ipv6.tp.dst,
+ tcp_key->tcp_dst, is_mask);
+ }
+ attrs &= ~(1 << OVS_KEY_ATTR_TCP);
+ }
+
+ if (attrs & (1 << OVS_KEY_ATTR_UDP)) {
+ const struct ovs_key_udp *udp_key;
+
+ udp_key = nla_data(a[OVS_KEY_ATTR_UDP]);
+ if (orig_attrs & (1 << OVS_KEY_ATTR_IPV4)) {
+ SW_FLOW_KEY_PUT(match, ipv4.tp.src,
+ udp_key->udp_src, is_mask);
+ SW_FLOW_KEY_PUT(match, ipv4.tp.dst,
+ udp_key->udp_dst, is_mask);
+ } else {
+ SW_FLOW_KEY_PUT(match, ipv6.tp.src,
+ udp_key->udp_src, is_mask);
+ SW_FLOW_KEY_PUT(match, ipv6.tp.dst,
+ udp_key->udp_dst, is_mask);
+ }
+ attrs &= ~(1 << OVS_KEY_ATTR_UDP);
+ }
+
+ if (attrs & (1 << OVS_KEY_ATTR_ICMP)) {
+ const struct ovs_key_icmp *icmp_key;
+
+ icmp_key = nla_data(a[OVS_KEY_ATTR_ICMP]);
+ SW_FLOW_KEY_PUT(match, ipv4.tp.src,
+ htons(icmp_key->icmp_type), is_mask);
+ SW_FLOW_KEY_PUT(match, ipv4.tp.dst,
+ htons(icmp_key->icmp_code), is_mask);
+ attrs &= ~(1 << OVS_KEY_ATTR_ICMP);
+ }
+
+ if (attrs & (1 << OVS_KEY_ATTR_ICMPV6)) {
+ const struct ovs_key_icmpv6 *icmpv6_key;
+
+ icmpv6_key = nla_data(a[OVS_KEY_ATTR_ICMPV6]);
+ SW_FLOW_KEY_PUT(match, ipv6.tp.src,
+ htons(icmpv6_key->icmpv6_type), is_mask);
+ SW_FLOW_KEY_PUT(match, ipv6.tp.dst,
+ htons(icmpv6_key->icmpv6_code), is_mask);
+ attrs &= ~(1 << OVS_KEY_ATTR_ICMPV6);
+ }
+
+ if (attrs & (1 << OVS_KEY_ATTR_ND)) {
+ const struct ovs_key_nd *nd_key;
+
+ nd_key = nla_data(a[OVS_KEY_ATTR_ND]);
+ SW_FLOW_KEY_MEMCPY(match, ipv6.nd.target,
+ nd_key->nd_target,
+ sizeof(match->key->ipv6.nd.target),
+ is_mask);
+ SW_FLOW_KEY_MEMCPY(match, ipv6.nd.sll,
+ nd_key->nd_sll, ETH_ALEN, is_mask);
+ SW_FLOW_KEY_MEMCPY(match, ipv6.nd.tll,
+ nd_key->nd_tll, ETH_ALEN, is_mask);
+ attrs &= ~(1 << OVS_KEY_ATTR_ND);
+ }
+
+ if (attrs != 0)
+ return -EINVAL;
+
+ return 0;
+}
+
+/**
+ * ovs_match_from_nlattrs - parses Netlink attributes into a flow key and
+ * mask. In case the 'mask' is NULL, the flow is treated as exact match
+ * flow. Otherwise, it is treated as a wildcarded flow, except the mask
+ * does not include any don't care bit.
+ * @match: receives the extracted flow match information.
+ * @key: Netlink attribute holding nested %OVS_KEY_ATTR_* Netlink attribute
+ * sequence. The fields should of the packet that triggered the creation
+ * of this flow.
+ * @mask: Optional. Netlink attribute holding nested %OVS_KEY_ATTR_* Netlink
+ * attribute specifies the mask field of the wildcarded flow.
+ */
+int ovs_match_from_nlattrs(struct sw_flow_match *match,
+ const struct nlattr *key,
+ const struct nlattr *mask)
+{
+ const struct nlattr *a[OVS_KEY_ATTR_MAX + 1];
+ const struct nlattr *encap;
+ u64 key_attrs = 0;
+ u64 mask_attrs = 0;
+ bool encap_valid = false;
+ int err;
+
+ err = parse_flow_nlattrs(key, a, &key_attrs);
+ if (err)
+ return err;
+
+ if ((key_attrs & (1 << OVS_KEY_ATTR_ETHERNET)) &&
+ (key_attrs & (1 << OVS_KEY_ATTR_ETHERTYPE)) &&
+ (nla_get_be16(a[OVS_KEY_ATTR_ETHERTYPE]) == htons(ETH_P_8021Q))) {
+ __be16 tci;
+
+ if (!((key_attrs & (1 << OVS_KEY_ATTR_VLAN)) &&
+ (key_attrs & (1 << OVS_KEY_ATTR_ENCAP)))) {
+ OVS_NLERR("Invalid Vlan frame.\n");
return -EINVAL;
- swkey->ip.proto = ntohs(arp_key->arp_op);
- memcpy(swkey->ipv4.arp.sha, arp_key->arp_sha, ETH_ALEN);
- memcpy(swkey->ipv4.arp.tha, arp_key->arp_tha, ETH_ALEN);
+ }
+
+ key_attrs &= ~(1 << OVS_KEY_ATTR_ETHERTYPE);
+ tci = nla_get_be16(a[OVS_KEY_ATTR_VLAN]);
+ encap = a[OVS_KEY_ATTR_ENCAP];
+ key_attrs &= ~(1 << OVS_KEY_ATTR_ENCAP);
+ encap_valid = true;
+
+ if (tci & htons(VLAN_TAG_PRESENT)) {
+ err = parse_flow_nlattrs(encap, a, &key_attrs);
+ if (err)
+ return err;
+ } else if (!tci) {
+ /* Corner case for truncated 802.1Q header. */
+ if (nla_len(encap)) {
+ OVS_NLERR("Truncated 802.1Q header has non-zero encap attribute.\n");
+ return -EINVAL;
+ }
+ } else {
+ OVS_NLERR("Encap attribute is set for a non-VLAN frame.\n");
+ return -EINVAL;
+ }
}
- if (attrs)
+ err = ovs_key_from_nlattrs(match, key_attrs, a, false);
+ if (err)
+ return err;
+
+ if (mask) {
+ err = parse_flow_mask_nlattrs(mask, a, &mask_attrs);
+ if (err)
+ return err;
+
+ if (mask_attrs & 1ULL << OVS_KEY_ATTR_ENCAP) {
+ __be16 eth_type = 0;
+ __be16 tci = 0;
+
+ if (!encap_valid) {
+ OVS_NLERR("Encap mask attribute is set for non-VLAN frame.\n");
+ return -EINVAL;
+ }
+
+ mask_attrs &= ~(1 << OVS_KEY_ATTR_ENCAP);
+ if (a[OVS_KEY_ATTR_ETHERTYPE])
+ eth_type = nla_get_be16(a[OVS_KEY_ATTR_ETHERTYPE]);
+
+ if (eth_type == htons(0xffff)) {
+ mask_attrs &= ~(1 << OVS_KEY_ATTR_ETHERTYPE);
+ encap = a[OVS_KEY_ATTR_ENCAP];
+ err = parse_flow_mask_nlattrs(encap, a, &mask_attrs);
+ } else {
+ OVS_NLERR("VLAN frames must have an exact match on the TPID (mask=%x).\n",
+ ntohs(eth_type));
+ return -EINVAL;
+ }
+
+ if (a[OVS_KEY_ATTR_VLAN])
+ tci = nla_get_be16(a[OVS_KEY_ATTR_VLAN]);
+
+ if (!(tci & htons(VLAN_TAG_PRESENT))) {
+ OVS_NLERR("VLAN tag present bit must have an exact match (tci_mask=%x).\n", ntohs(tci));
+ return -EINVAL;
+ }
+ }
+
+ err = ovs_key_from_nlattrs(match, mask_attrs, a, true);
+ if (err)
+ return err;
+ } else {
+ /* Populate exact match flow's key mask. */
+ if (match->mask)
+ ovs_sw_flow_mask_set(match->mask, &match->range, 0xff);
+ }
+
+ if (!ovs_match_validate(match, key_attrs, mask_attrs))
return -EINVAL;
- *key_lenp = key_len;
return 0;
}
@@ -1255,7 +1662,6 @@ int ovs_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp,
/**
* ovs_flow_metadata_from_nlattrs - parses Netlink attributes into a flow key.
* @flow: Receives extracted in_port, priority, tun_key and skb_mark.
- * @key_len: Length of key in @flow. Used for calculating flow hash.
* @attr: Netlink attribute holding nested %OVS_KEY_ATTR_* Netlink attribute
* sequence.
*
@@ -1264,102 +1670,100 @@ int ovs_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp,
* get the metadata, that is, the parts of the flow key that cannot be
* extracted from the packet itself.
*/
-int ovs_flow_metadata_from_nlattrs(struct sw_flow *flow, int key_len,
- const struct nlattr *attr)
+
+int ovs_flow_metadata_from_nlattrs(struct sw_flow *flow,
+ const struct nlattr *attr)
{
struct ovs_key_ipv4_tunnel *tun_key = &flow->key.tun_key;
- const struct nlattr *nla;
- int rem;
+ const struct nlattr *a[OVS_KEY_ATTR_MAX + 1];
+ u64 attrs = 0;
+ int err;
+ struct sw_flow_match match;
flow->key.phy.in_port = DP_MAX_PORTS;
flow->key.phy.priority = 0;
flow->key.phy.skb_mark = 0;
memset(tun_key, 0, sizeof(flow->key.tun_key));
- nla_for_each_nested(nla, attr, rem) {
- int type = nla_type(nla);
-
- if (type <= OVS_KEY_ATTR_MAX && ovs_key_lens[type] > 0) {
- int err;
-
- if (nla_len(nla) != ovs_key_lens[type])
- return -EINVAL;
-
- switch (type) {
- case OVS_KEY_ATTR_PRIORITY:
- flow->key.phy.priority = nla_get_u32(nla);
- break;
-
- case OVS_KEY_ATTR_TUNNEL:
- err = ovs_ipv4_tun_from_nlattr(nla, tun_key);
- if (err)
- return err;
- break;
-
- case OVS_KEY_ATTR_IN_PORT:
- if (nla_get_u32(nla) >= DP_MAX_PORTS)
- return -EINVAL;
- flow->key.phy.in_port = nla_get_u32(nla);
- break;
-
- case OVS_KEY_ATTR_SKB_MARK:
- flow->key.phy.skb_mark = nla_get_u32(nla);
- break;
- }
- }
- }
- if (rem)
+ err = parse_flow_nlattrs(attr, a, &attrs);
+ if (err)
return -EINVAL;
- flow->hash = ovs_flow_hash(&flow->key,
- flow_key_start(&flow->key), key_len);
+ memset(&match, 0, sizeof(match));
+ match.key = &flow->key;
+
+ err = metadata_from_nlattrs(&match, &attrs, a, false);
+ if (err)
+ return err;
return 0;
}
-int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
+int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey,
+ const struct sw_flow_key *output, struct sk_buff *skb)
{
struct ovs_key_ethernet *eth_key;
struct nlattr *nla, *encap;
+ bool is_mask = (swkey != output);
- if (swkey->phy.priority &&
- nla_put_u32(skb, OVS_KEY_ATTR_PRIORITY, swkey->phy.priority))
+ if (nla_put_u32(skb, OVS_KEY_ATTR_PRIORITY, output->phy.priority))
goto nla_put_failure;
- if (swkey->tun_key.ipv4_dst &&
- ovs_ipv4_tun_to_nlattr(skb, &swkey->tun_key))
+ if ((swkey->tun_key.ipv4_dst || is_mask) &&
+ ovs_ipv4_tun_to_nlattr(skb, &swkey->tun_key, &output->tun_key))
goto nla_put_failure;
- if (swkey->phy.in_port != DP_MAX_PORTS &&
- nla_put_u32(skb, OVS_KEY_ATTR_IN_PORT, swkey->phy.in_port))
- goto nla_put_failure;
+ if (swkey->phy.in_port == DP_MAX_PORTS) {
+ if (is_mask && (output->phy.in_port == 0xffff))
+ if (nla_put_u32(skb, OVS_KEY_ATTR_IN_PORT, 0xffffffff))
+ goto nla_put_failure;
+ } else {
+ u16 upper_u16;
+ upper_u16 = !is_mask ? 0 : 0xffff;
- if (swkey->phy.skb_mark &&
- nla_put_u32(skb, OVS_KEY_ATTR_SKB_MARK, swkey->phy.skb_mark))
+ if (nla_put_u32(skb, OVS_KEY_ATTR_IN_PORT,
+ (upper_u16 << 16) | output->phy.in_port))
+ goto nla_put_failure;
+ }
+
+ if (nla_put_u32(skb, OVS_KEY_ATTR_SKB_MARK, output->phy.skb_mark))
goto nla_put_failure;
nla = nla_reserve(skb, OVS_KEY_ATTR_ETHERNET, sizeof(*eth_key));
if (!nla)
goto nla_put_failure;
+
eth_key = nla_data(nla);
- memcpy(eth_key->eth_src, swkey->eth.src, ETH_ALEN);
- memcpy(eth_key->eth_dst, swkey->eth.dst, ETH_ALEN);
+ memcpy(eth_key->eth_src, output->eth.src, ETH_ALEN);
+ memcpy(eth_key->eth_dst, output->eth.dst, ETH_ALEN);
if (swkey->eth.tci || swkey->eth.type == htons(ETH_P_8021Q)) {
- if (nla_put_be16(skb, OVS_KEY_ATTR_ETHERTYPE, htons(ETH_P_8021Q)) ||
- nla_put_be16(skb, OVS_KEY_ATTR_VLAN, swkey->eth.tci))
+ __be16 eth_type;
+ eth_type = !is_mask ? htons(ETH_P_8021Q) : htons(0xffff);
+ if (nla_put_be16(skb, OVS_KEY_ATTR_ETHERTYPE, eth_type) ||
+ nla_put_be16(skb, OVS_KEY_ATTR_VLAN, output->eth.tci))
goto nla_put_failure;
encap = nla_nest_start(skb, OVS_KEY_ATTR_ENCAP);
if (!swkey->eth.tci)
goto unencap;
- } else {
+ } else
encap = NULL;
- }
- if (swkey->eth.type == htons(ETH_P_802_2))
+ if (swkey->eth.type == htons(ETH_P_802_2)) {
+ /*
+ * Ethertype 802.2 is represented in the netlink with omitted
+ * OVS_KEY_ATTR_ETHERTYPE in the flow key attribute, and
+ * 0xffff in the mask attribute. Ethertype can also
+ * be wildcarded.
+ */
+ if (is_mask && output->eth.type)
+ if (nla_put_be16(skb, OVS_KEY_ATTR_ETHERTYPE,
+ output->eth.type))
+ goto nla_put_failure;
goto unencap;
+ }
- if (nla_put_be16(skb, OVS_KEY_ATTR_ETHERTYPE, swkey->eth.type))
+ if (nla_put_be16(skb, OVS_KEY_ATTR_ETHERTYPE, output->eth.type))
goto nla_put_failure;
if (swkey->eth.type == htons(ETH_P_IP)) {
@@ -1369,12 +1773,12 @@ int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
if (!nla)
goto nla_put_failure;
ipv4_key = nla_data(nla);
- ipv4_key->ipv4_src = swkey->ipv4.addr.src;
- ipv4_key->ipv4_dst = swkey->ipv4.addr.dst;
- ipv4_key->ipv4_proto = swkey->ip.proto;
- ipv4_key->ipv4_tos = swkey->ip.tos;
- ipv4_key->ipv4_ttl = swkey->ip.ttl;
- ipv4_key->ipv4_frag = swkey->ip.frag;
+ ipv4_key->ipv4_src = output->ipv4.addr.src;
+ ipv4_key->ipv4_dst = output->ipv4.addr.dst;
+ ipv4_key->ipv4_proto = output->ip.proto;
+ ipv4_key->ipv4_tos = output->ip.tos;
+ ipv4_key->ipv4_ttl = output->ip.ttl;
+ ipv4_key->ipv4_frag = output->ip.frag;
} else if (swkey->eth.type == htons(ETH_P_IPV6)) {
struct ovs_key_ipv6 *ipv6_key;
@@ -1382,15 +1786,15 @@ int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
if (!nla)
goto nla_put_failure;
ipv6_key = nla_data(nla);
- memcpy(ipv6_key->ipv6_src, &swkey->ipv6.addr.src,
+ memcpy(ipv6_key->ipv6_src, &output->ipv6.addr.src,
sizeof(ipv6_key->ipv6_src));
- memcpy(ipv6_key->ipv6_dst, &swkey->ipv6.addr.dst,
+ memcpy(ipv6_key->ipv6_dst, &output->ipv6.addr.dst,
sizeof(ipv6_key->ipv6_dst));
- ipv6_key->ipv6_label = swkey->ipv6.label;
- ipv6_key->ipv6_proto = swkey->ip.proto;
- ipv6_key->ipv6_tclass = swkey->ip.tos;
- ipv6_key->ipv6_hlimit = swkey->ip.ttl;
- ipv6_key->ipv6_frag = swkey->ip.frag;
+ ipv6_key->ipv6_label = output->ipv6.label;
+ ipv6_key->ipv6_proto = output->ip.proto;
+ ipv6_key->ipv6_tclass = output->ip.tos;
+ ipv6_key->ipv6_hlimit = output->ip.ttl;
+ ipv6_key->ipv6_frag = output->ip.frag;
} else if (swkey->eth.type == htons(ETH_P_ARP) ||
swkey->eth.type == htons(ETH_P_RARP)) {
struct ovs_key_arp *arp_key;
@@ -1400,11 +1804,11 @@ int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
goto nla_put_failure;
arp_key = nla_data(nla);
memset(arp_key, 0, sizeof(struct ovs_key_arp));
- arp_key->arp_sip = swkey->ipv4.addr.src;
- arp_key->arp_tip = swkey->ipv4.addr.dst;
- arp_key->arp_op = htons(swkey->ip.proto);
- memcpy(arp_key->arp_sha, swkey->ipv4.arp.sha, ETH_ALEN);
- memcpy(arp_key->arp_tha, swkey->ipv4.arp.tha, ETH_ALEN);
+ arp_key->arp_sip = output->ipv4.addr.src;
+ arp_key->arp_tip = output->ipv4.addr.dst;
+ arp_key->arp_op = htons(output->ip.proto);
+ memcpy(arp_key->arp_sha, output->ipv4.arp.sha, ETH_ALEN);
+ memcpy(arp_key->arp_tha, output->ipv4.arp.tha, ETH_ALEN);
}
if ((swkey->eth.type == htons(ETH_P_IP) ||
@@ -1419,11 +1823,11 @@ int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
goto nla_put_failure;
tcp_key = nla_data(nla);
if (swkey->eth.type == htons(ETH_P_IP)) {
- tcp_key->tcp_src = swkey->ipv4.tp.src;
- tcp_key->tcp_dst = swkey->ipv4.tp.dst;
+ tcp_key->tcp_src = output->ipv4.tp.src;
+ tcp_key->tcp_dst = output->ipv4.tp.dst;
} else if (swkey->eth.type == htons(ETH_P_IPV6)) {
- tcp_key->tcp_src = swkey->ipv6.tp.src;
- tcp_key->tcp_dst = swkey->ipv6.tp.dst;
+ tcp_key->tcp_src = output->ipv6.tp.src;
+ tcp_key->tcp_dst = output->ipv6.tp.dst;
}
} else if (swkey->ip.proto == IPPROTO_UDP) {
struct ovs_key_udp *udp_key;
@@ -1433,11 +1837,11 @@ int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
goto nla_put_failure;
udp_key = nla_data(nla);
if (swkey->eth.type == htons(ETH_P_IP)) {
- udp_key->udp_src = swkey->ipv4.tp.src;
- udp_key->udp_dst = swkey->ipv4.tp.dst;
+ udp_key->udp_src = output->ipv4.tp.src;
+ udp_key->udp_dst = output->ipv4.tp.dst;
} else if (swkey->eth.type == htons(ETH_P_IPV6)) {
- udp_key->udp_src = swkey->ipv6.tp.src;
- udp_key->udp_dst = swkey->ipv6.tp.dst;
+ udp_key->udp_src = output->ipv6.tp.src;
+ udp_key->udp_dst = output->ipv6.tp.dst;
}
} else if (swkey->eth.type == htons(ETH_P_IP) &&
swkey->ip.proto == IPPROTO_ICMP) {
@@ -1447,8 +1851,8 @@ int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
if (!nla)
goto nla_put_failure;
icmp_key = nla_data(nla);
- icmp_key->icmp_type = ntohs(swkey->ipv4.tp.src);
- icmp_key->icmp_code = ntohs(swkey->ipv4.tp.dst);
+ icmp_key->icmp_type = ntohs(output->ipv4.tp.src);
+ icmp_key->icmp_code = ntohs(output->ipv4.tp.dst);
} else if (swkey->eth.type == htons(ETH_P_IPV6) &&
swkey->ip.proto == IPPROTO_ICMPV6) {
struct ovs_key_icmpv6 *icmpv6_key;
@@ -1458,8 +1862,8 @@ int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
if (!nla)
goto nla_put_failure;
icmpv6_key = nla_data(nla);
- icmpv6_key->icmpv6_type = ntohs(swkey->ipv6.tp.src);
- icmpv6_key->icmpv6_code = ntohs(swkey->ipv6.tp.dst);
+ icmpv6_key->icmpv6_type = ntohs(output->ipv6.tp.src);
+ icmpv6_key->icmpv6_code = ntohs(output->ipv6.tp.dst);
if (icmpv6_key->icmpv6_type == NDISC_NEIGHBOUR_SOLICITATION ||
icmpv6_key->icmpv6_type == NDISC_NEIGHBOUR_ADVERTISEMENT) {
@@ -1469,10 +1873,10 @@ int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
if (!nla)
goto nla_put_failure;
nd_key = nla_data(nla);
- memcpy(nd_key->nd_target, &swkey->ipv6.nd.target,
+ memcpy(nd_key->nd_target, &output->ipv6.nd.target,
sizeof(nd_key->nd_target));
- memcpy(nd_key->nd_sll, swkey->ipv6.nd.sll, ETH_ALEN);
- memcpy(nd_key->nd_tll, swkey->ipv6.nd.tll, ETH_ALEN);
+ memcpy(nd_key->nd_sll, output->ipv6.nd.sll, ETH_ALEN);
+ memcpy(nd_key->nd_tll, output->ipv6.nd.tll, ETH_ALEN);
}
}
}
@@ -1504,3 +1908,84 @@ void ovs_flow_exit(void)
{
kmem_cache_destroy(flow_cache);
}
+
+struct sw_flow_mask *ovs_sw_flow_mask_alloc(void)
+{
+ struct sw_flow_mask *mask;
+
+ mask = kmalloc(sizeof(*mask), GFP_KERNEL);
+ if (mask)
+ mask->ref_count = 0;
+
+ return mask;
+}
+
+void ovs_sw_flow_mask_add_ref(struct sw_flow_mask *mask)
+{
+ mask->ref_count++;
+}
+
+void ovs_sw_flow_mask_del_ref(struct sw_flow_mask *mask, bool deferred)
+{
+ if (!mask)
+ return;
+
+ BUG_ON(!mask->ref_count);
+ mask->ref_count--;
+
+ if (!mask->ref_count) {
+ list_del_rcu(&mask->list);
+ if (deferred)
+ kfree_rcu(mask, rcu);
+ else
+ kfree(mask);
+ }
+}
+
+static bool ovs_sw_flow_mask_equal(const struct sw_flow_mask *a,
+ const struct sw_flow_mask *b)
+{
+ u8 *a_ = (u8 *)&a->key + a->range.start;
+ u8 *b_ = (u8 *)&b->key + b->range.start;
+
+ return (a->range.end == b->range.end)
+ && (a->range.start == b->range.start)
+ && (memcmp(a_, b_, ovs_sw_flow_mask_actual_size(a)) == 0);
+}
+
+struct sw_flow_mask *ovs_sw_flow_mask_find(const struct flow_table *tbl,
+ const struct sw_flow_mask *mask)
+{
+ struct list_head *ml;
+
+ list_for_each(ml, tbl->mask_list) {
+ struct sw_flow_mask *m;
+ m = container_of(ml, struct sw_flow_mask, list);
+ if (ovs_sw_flow_mask_equal(mask, m))
+ return m;
+ }
+
+ return NULL;
+}
+
+/**
+ * add a new mask into the mask list.
+ * The caller needs to make sure that 'mask' is not the same
+ * as any masks that are already on the list.
+ */
+void ovs_sw_flow_mask_insert(struct flow_table *tbl, struct sw_flow_mask *mask)
+{
+ list_add_rcu(&mask->list, tbl->mask_list);
+}
+
+/**
+ * Set 'range' fields in the mask to the value of 'val'.
+ */
+static void ovs_sw_flow_mask_set(struct sw_flow_mask *mask,
+ struct sw_flow_key_range *range, u8 val)
+{
+ u8 *m = (u8 *)&mask->key + range->start;
+
+ mask->range = *range;
+ memset(m, val, ovs_sw_flow_mask_size_roundup(mask));
+}
diff --git a/net/openvswitch/flow.h b/net/openvswitch/flow.h
index 66ef722..9674e45 100644
--- a/net/openvswitch/flow.h
+++ b/net/openvswitch/flow.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2007-2011 Nicira, Inc.
+ * Copyright (c) 2007-2013 Nicira, Inc.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of version 2 of the GNU General Public
@@ -33,6 +33,8 @@
#include <net/inet_ecn.h>
struct sk_buff;
+struct sw_flow_mask;
+struct flow_table;
struct sw_flow_actions {
struct rcu_head rcu;
@@ -131,6 +133,8 @@ struct sw_flow {
u32 hash;
struct sw_flow_key key;
+ struct sw_flow_key unmasked_key;
+ struct sw_flow_mask *mask;
struct sw_flow_actions __rcu *sf_acts;
spinlock_t lock; /* Lock for values below. */
@@ -140,6 +144,25 @@ struct sw_flow {
u8 tcp_flags; /* Union of seen TCP flags. */
};
+struct sw_flow_key_range {
+ size_t start;
+ size_t end;
+};
+
+static inline u16 ovs_sw_flow_key_range_actual_size(const struct sw_flow_key_range *range)
+{
+ return range->end - range->start;
+}
+
+struct sw_flow_match {
+ struct sw_flow_key *key;
+ struct sw_flow_key_range range;
+ struct sw_flow_mask *mask;
+};
+
+void ovs_match_init(struct sw_flow_match *match,
+ struct sw_flow_key *key, struct sw_flow_mask *mask);
+
struct arp_eth_header {
__be16 ar_hrd; /* format of hardware address */
__be16 ar_pro; /* format of protocol address */
@@ -159,21 +182,21 @@ void ovs_flow_exit(void);
struct sw_flow *ovs_flow_alloc(void);
void ovs_flow_deferred_free(struct sw_flow *);
-void ovs_flow_free(struct sw_flow *flow);
+void ovs_flow_free(struct sw_flow *, bool deferred);
struct sw_flow_actions *ovs_flow_actions_alloc(int actions_len);
void ovs_flow_deferred_free_acts(struct sw_flow_actions *);
-int ovs_flow_extract(struct sk_buff *, u16 in_port, struct sw_flow_key *,
- int *key_lenp);
+int ovs_flow_extract(struct sk_buff *, u16 in_port, struct sw_flow_key *);
void ovs_flow_used(struct sw_flow *, struct sk_buff *);
u64 ovs_flow_used_time(unsigned long flow_jiffies);
-
-int ovs_flow_to_nlattrs(const struct sw_flow_key *, struct sk_buff *);
-int ovs_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp,
+int ovs_flow_to_nlattrs(const struct sw_flow_key *,
+ const struct sw_flow_key *, struct sk_buff *);
+int ovs_match_from_nlattrs(struct sw_flow_match *match,
+ const struct nlattr *,
const struct nlattr *);
-int ovs_flow_metadata_from_nlattrs(struct sw_flow *flow, int key_len,
- const struct nlattr *attr);
+int ovs_flow_metadata_from_nlattrs(struct sw_flow *flow,
+ const struct nlattr *attr);
#define MAX_ACTIONS_BUFSIZE (32 * 1024)
#define TBL_MIN_BUCKETS 1024
@@ -182,6 +205,7 @@ struct flow_table {
struct flex_array *buckets;
unsigned int count, n_buckets;
struct rcu_head rcu;
+ struct list_head *mask_list;
int node_ver;
u32 hash_seed;
bool keep_flows;
@@ -197,22 +221,56 @@ static inline int ovs_flow_tbl_need_to_expand(struct flow_table *table)
return (table->count > table->n_buckets);
}
-struct sw_flow *ovs_flow_tbl_lookup(struct flow_table *table,
- struct sw_flow_key *key, int len);
-void ovs_flow_tbl_destroy(struct flow_table *table);
-void ovs_flow_tbl_deferred_destroy(struct flow_table *table);
+struct sw_flow *ovs_flow_lookup(struct flow_table *,
+ const struct sw_flow_key *);
+struct sw_flow *ovs_flow_lookup_unmasked_key(struct flow_table *table,
+ struct sw_flow_match *match);
+
+void ovs_flow_tbl_destroy(struct flow_table *table, bool deferred);
struct flow_table *ovs_flow_tbl_alloc(int new_size);
struct flow_table *ovs_flow_tbl_expand(struct flow_table *table);
struct flow_table *ovs_flow_tbl_rehash(struct flow_table *table);
-void ovs_flow_tbl_insert(struct flow_table *table, struct sw_flow *flow,
- struct sw_flow_key *key, int key_len);
-void ovs_flow_tbl_remove(struct flow_table *table, struct sw_flow *flow);
-struct sw_flow *ovs_flow_tbl_next(struct flow_table *table, u32 *bucket, u32 *idx);
+void ovs_flow_insert(struct flow_table *table, struct sw_flow *flow);
+void ovs_flow_remove(struct flow_table *table, struct sw_flow *flow);
+
+struct sw_flow *ovs_flow_dump_next(struct flow_table *table, u32 *bucket, u32 *idx);
extern const int ovs_key_lens[OVS_KEY_ATTR_MAX + 1];
int ovs_ipv4_tun_from_nlattr(const struct nlattr *attr,
- struct ovs_key_ipv4_tunnel *tun_key);
+ struct sw_flow_match *match, bool is_mask);
int ovs_ipv4_tun_to_nlattr(struct sk_buff *skb,
- const struct ovs_key_ipv4_tunnel *tun_key);
+ const struct ovs_key_ipv4_tunnel *tun_key,
+ const struct ovs_key_ipv4_tunnel *output);
+
+bool ovs_flow_cmp_unmasked_key(const struct sw_flow *flow,
+ const struct sw_flow_key *key, int key_len);
+
+struct sw_flow_mask {
+ int ref_count;
+ struct rcu_head rcu;
+ struct list_head list;
+ struct sw_flow_key_range range;
+ struct sw_flow_key key;
+};
+
+static inline u16
+ovs_sw_flow_mask_actual_size(const struct sw_flow_mask *mask)
+{
+ return ovs_sw_flow_key_range_actual_size(&mask->range);
+}
+
+static inline u16
+ovs_sw_flow_mask_size_roundup(const struct sw_flow_mask *mask)
+{
+ return roundup(ovs_sw_flow_mask_actual_size(mask), sizeof(u32));
+}
+struct sw_flow_mask *ovs_sw_flow_mask_alloc(void);
+void ovs_sw_flow_mask_add_ref(struct sw_flow_mask *);
+void ovs_sw_flow_mask_del_ref(struct sw_flow_mask *, bool deferred);
+void ovs_sw_flow_mask_insert(struct flow_table *, struct sw_flow_mask *);
+struct sw_flow_mask *ovs_sw_flow_mask_find(const struct flow_table *,
+ const struct sw_flow_mask *);
+void ovs_flow_key_mask(struct sw_flow_key *dst, const struct sw_flow_key *src,
+ const struct sw_flow_mask *mask);
#endif /* flow.h */
--
1.8.1.2
^ permalink raw reply related
* [PATCH net-next 06/11] openvswitch: check CONFIG_OPENVSWITCH_GRE in makefile
From: Jesse Gross @ 2013-08-27 20:20 UTC (permalink / raw)
To: David Miller; +Cc: netdev, dev, Cong Wang, Pravin B Shelar, Jesse Gross
In-Reply-To: <1377634848-34327-1-git-send-email-jesse@nicira.com>
From: Cong Wang <amwang@redhat.com>
Cc: Jesse Gross <jesse@nicira.com>
Cc: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: Cong Wang <amwang@redhat.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>
---
net/openvswitch/Makefile | 5 ++++-
net/openvswitch/vport-gre.c | 3 ---
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/net/openvswitch/Makefile b/net/openvswitch/Makefile
index 82e4ee5..ea36e99 100644
--- a/net/openvswitch/Makefile
+++ b/net/openvswitch/Makefile
@@ -10,10 +10,13 @@ openvswitch-y := \
dp_notify.o \
flow.o \
vport.o \
- vport-gre.o \
vport-internal_dev.o \
vport-netdev.o
ifneq ($(CONFIG_OPENVSWITCH_VXLAN),)
openvswitch-y += vport-vxlan.o
endif
+
+ifneq ($(CONFIG_OPENVSWITCH_GRE),)
+openvswitch-y += vport-gre.o
+endif
diff --git a/net/openvswitch/vport-gre.c b/net/openvswitch/vport-gre.c
index 493e977..21d5073 100644
--- a/net/openvswitch/vport-gre.c
+++ b/net/openvswitch/vport-gre.c
@@ -16,7 +16,6 @@
* 02110-1301, USA
*/
-#ifdef CONFIG_OPENVSWITCH_GRE
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
#include <linux/if.h>
@@ -271,5 +270,3 @@ const struct vport_ops ovs_gre_vport_ops = {
.get_name = gre_get_name,
.send = gre_tnl_send,
};
-
-#endif /* OPENVSWITCH_GRE */
--
1.8.1.2
^ permalink raw reply related
* [PATCH net-next 04/11] openvswitch:: link upper device for port devices
From: Jesse Gross @ 2013-08-27 20:20 UTC (permalink / raw)
To: David Miller; +Cc: netdev, dev, Jiri Pirko, Jesse Gross
In-Reply-To: <1377634848-34327-1-git-send-email-jesse@nicira.com>
From: Jiri Pirko <jiri@resnulli.us>
Link upper device properly. That will make IFLA_MASTER filled up.
Set the master to port 0 of the datapath under which the port belongs.
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: Jesse Gross <jesse@nicira.com>
---
net/openvswitch/vport-netdev.c | 20 +++++++++++++++++++-
1 file changed, 19 insertions(+), 1 deletion(-)
diff --git a/net/openvswitch/vport-netdev.c b/net/openvswitch/vport-netdev.c
index 5982f3f..09d93c1 100644
--- a/net/openvswitch/vport-netdev.c
+++ b/net/openvswitch/vport-netdev.c
@@ -25,6 +25,7 @@
#include <linux/llc.h>
#include <linux/rtnetlink.h>
#include <linux/skbuff.h>
+#include <linux/openvswitch.h>
#include <net/llc.h>
@@ -74,6 +75,15 @@ static rx_handler_result_t netdev_frame_hook(struct sk_buff **pskb)
return RX_HANDLER_CONSUMED;
}
+static struct net_device *get_dpdev(struct datapath *dp)
+{
+ struct vport *local;
+
+ local = ovs_vport_ovsl(dp, OVSP_LOCAL);
+ BUG_ON(!local);
+ return netdev_vport_priv(local)->dev;
+}
+
static struct vport *netdev_create(const struct vport_parms *parms)
{
struct vport *vport;
@@ -103,10 +113,15 @@ static struct vport *netdev_create(const struct vport_parms *parms)
}
rtnl_lock();
+ err = netdev_master_upper_dev_link(netdev_vport->dev,
+ get_dpdev(vport->dp));
+ if (err)
+ goto error_unlock;
+
err = netdev_rx_handler_register(netdev_vport->dev, netdev_frame_hook,
vport);
if (err)
- goto error_unlock;
+ goto error_master_upper_dev_unlink;
dev_set_promiscuity(netdev_vport->dev, 1);
netdev_vport->dev->priv_flags |= IFF_OVS_DATAPATH;
@@ -114,6 +129,8 @@ static struct vport *netdev_create(const struct vport_parms *parms)
return vport;
+error_master_upper_dev_unlink:
+ netdev_upper_dev_unlink(netdev_vport->dev, get_dpdev(vport->dp));
error_unlock:
rtnl_unlock();
error_put:
@@ -140,6 +157,7 @@ static void netdev_destroy(struct vport *vport)
rtnl_lock();
netdev_vport->dev->priv_flags &= ~IFF_OVS_DATAPATH;
netdev_rx_handler_unregister(netdev_vport->dev);
+ netdev_upper_dev_unlink(netdev_vport->dev, get_dpdev(vport->dp));
dev_set_promiscuity(netdev_vport->dev, -1);
rtnl_unlock();
--
1.8.1.2
^ permalink raw reply related
* [PATCH net-next 03/11] openvswitch: Use non rcu hlist_del() flow table entry.
From: Jesse Gross @ 2013-08-27 20:20 UTC (permalink / raw)
To: David Miller; +Cc: netdev, dev, Pravin B Shelar, Jesse Gross
In-Reply-To: <1377634848-34327-1-git-send-email-jesse@nicira.com>
From: Pravin B Shelar <pshelar@nicira.com>
Flow table destroy is done in rcu call-back context. Therefore
there is no need to use rcu variant of hlist_del().
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>
---
net/openvswitch/flow.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/openvswitch/flow.c b/net/openvswitch/flow.c
index 1aa84dc..fca2825 100644
--- a/net/openvswitch/flow.c
+++ b/net/openvswitch/flow.c
@@ -302,7 +302,7 @@ void ovs_flow_tbl_destroy(struct flow_table *table)
int ver = table->node_ver;
hlist_for_each_entry_safe(flow, n, head, hash_node[ver]) {
- hlist_del_rcu(&flow->hash_node[ver]);
+ hlist_del(&flow->hash_node[ver]);
ovs_flow_free(flow);
}
}
--
1.8.1.2
^ permalink raw reply related
* [PATCH net-next 02/11] openvswitch: Use RCU lock for dp dump operation.
From: Jesse Gross @ 2013-08-27 20:20 UTC (permalink / raw)
To: David Miller; +Cc: netdev, dev, Pravin B Shelar, Jesse Gross
In-Reply-To: <1377634848-34327-1-git-send-email-jesse@nicira.com>
From: Pravin B Shelar <pshelar@nicira.com>
RCUfy dp-dump operation which is already read-only. This
makes all ovs dump operations lockless.
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: Jesse Gross <jesse@nicira.com>
---
net/openvswitch/datapath.c | 13 +++++++------
1 file changed, 7 insertions(+), 6 deletions(-)
diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index e6fb866..9d97ef3 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -951,9 +951,10 @@ static struct genl_ops dp_packet_genl_ops[] = {
static void get_dp_stats(struct datapath *dp, struct ovs_dp_stats *stats)
{
+ struct flow_table *table;
int i;
- struct flow_table *table = ovsl_dereference(dp->table);
+ table = rcu_dereference_check(dp->table, lockdep_ovsl_is_held());
stats->n_flows = ovs_flow_tbl_count(table);
stats->n_hit = stats->n_missed = stats->n_lost = 0;
@@ -1665,7 +1666,7 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, struct genl_info *info)
goto err_destroy_local_port;
ovs_net = net_generic(ovs_dp_get_net(dp), ovs_net_id);
- list_add_tail(&dp->list_node, &ovs_net->dps);
+ list_add_tail_rcu(&dp->list_node, &ovs_net->dps);
ovs_unlock();
@@ -1703,7 +1704,7 @@ static void __dp_destroy(struct datapath *dp)
ovs_dp_detach_port(vport);
}
- list_del(&dp->list_node);
+ list_del_rcu(&dp->list_node);
/* OVSP_LOCAL is datapath internal port. We need to make sure that
* all port in datapath are destroyed first before freeing datapath.
@@ -1808,8 +1809,8 @@ static int ovs_dp_cmd_dump(struct sk_buff *skb, struct netlink_callback *cb)
int skip = cb->args[0];
int i = 0;
- ovs_lock();
- list_for_each_entry(dp, &ovs_net->dps, list_node) {
+ rcu_read_lock();
+ list_for_each_entry_rcu(dp, &ovs_net->dps, list_node) {
if (i >= skip &&
ovs_dp_cmd_fill_info(dp, skb, NETLINK_CB(cb->skb).portid,
cb->nlh->nlmsg_seq, NLM_F_MULTI,
@@ -1817,7 +1818,7 @@ static int ovs_dp_cmd_dump(struct sk_buff *skb, struct netlink_callback *cb)
break;
i++;
}
- ovs_unlock();
+ rcu_read_unlock();
cb->args[0] = i;
--
1.8.1.2
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox